Cheng Emag

DAVID K. CHENG SYRACUSE UNIVERSITY

A vv ADDISON-WESLEY PUBLISHING COMPANY

Reading, Massachusetts Menlo Park, California London Amsterdam Don Mills, Ontario Sydney

PARTH

Placed Image

DAVID K. CHENG SYRACUSE UNIVERSITY

A , vv ADDISON-WESLEY PUBLISHING COMPANY

I 1 Reading, Massachusetts Menlo Park, California

m on don' Amsterdam Don Mills, Ontario' Sydney

I

This book is in the API)ISON7WESLEY SERIES IN ELERRICAL ENGINEERING , .

SPONSORING EDIT& Tom Robbins PRODUCTION EDITOR: Marilee Sorotskin

TEXT DESIGNER: Mednda Grbsser ILLUSTRATOR: Dick Morton , - COVER DESIGNER AND ILLUSTRATOR: Richard H a n y s ART COORDINATOR: Dick Morton i.: PRODUCTION MANAGER: Herbert Nolan

The text of this book was composed in Times Roman by Syntax International.

Library of Congress Cataloging in Publication Data

Cheng, David K. avid-~(euni dqte- Field and wave e ~ e c t r o ~ ~ e t i ~ s .

Bibliography: p. I

1. Electromagnetism. 2. Fiqld th :ory (Physics) I. Title. I - QC760. C48 < 530.1'41 . 81-12749 ISBN 0-201-01239-1 AACR2 .

3 I I

I

COPY%^^ O 1983 by Addison-W+ey Publishing Company, Inc. rights reserved. NO part of @is publication may be reproduqd, btored in a retrieval system, or tnns-

m,ltted. in any, form or by any means. electmnic, mechanica~. photocopying, recordmg, or othenu~se, without the prior written permission df the publisher. Prmted m the Uolted States of Amenca. Published simultaneously in Canada. ,

ISBN 0-201-01239-1 ABCDEFGHIJ-AL-89876543

I

1-1 Introduction 1-2 The electromagnetic model 1-3 SI units and universal constants

Review questions

Introduction Vector addition and subtraction Products of vectors 2-3.1 Scalar or dot product 2-3.2 Vector or cross product 2-3.3 Product of three vectors Orthogonal coordinate systems 2-4.1 Cartesian coordinates 2-4.2 Cylindrical coordinates 2-4.3 Spherical coordinates Gradient of a scalar field Divergence of a vector field Divergence theorein Curl of a vector field Stokes's theorem

2-10 Two null identities 2-10.1 Identity I 2-10.1 Identity I1

2-1 1 Helmholtz's theorem Review questions . Problems

3-1 Introduction 3-2 Fundamental postulates of electrostatics in

free space - I

J-2 Coulonb's law 3 ?.: Electric fitla ciue LO a system of discrete charges 3-3.2 Electric field due to a contipuous distribution of charge

3-4 Gauss's law and applications 3-5 Electric potential I

3-5.1 Electric potential due to a charge distribution 3-6 Condugtors in static electric field 3-7 Dielectrics in static electric field

3-7.1 Equivalent charge distributions of polarized dielectrics

3-8 Electric flux density and dielectric constant 3-8.1 ~ielectric'strength

3-9 Boundary conditions for electrostatic fields 3-10 Capacitance and ~apacitors

3 - 10.1 Series aqd parallel connections of capacitors

3-1 1 Electrostatic enerpy and forces . , 3- 11.1 ~lectrost'atic energy in terms of field quantities 3- 11.2 Electrostatic forces Review. questions Problems 6 , I

\- :

2 .

Solution of Electrostati~ Problems . . 4-1 Introduction i

4-2 Poisson's and La~lace's equations 4-3 Uniqueness of eleptrgstatic solutions i 4-4 Method of imageg' : : I

4-4.1 Point charge and conducting planes 4-4.2 Line charge and parallel . I

conducting cylindpr . 4-4.3 Point charge aqd conducting sbhera,

1

4-5 Boundary-value problems in Cartesian coordinates

, 4-6 Boundary-value problems in cylindrical coordinates

4-7 Boundary-value problems in spherical coordinates Review questions ~roblems .

Steady Electric Currents

5-1 Introduction 5-2 Current density and Ohm's law 5-3 Electromotive force and Kirchhoff's

voltage law 5-4 Equation of continuity and Kirchhoff's current law 5-5 Power dissipation and Joulc's law 5-6 Boundary conditions for current density, 5-7 Resistance calculations

Review questions Problems

1

Static Magnetic Fields , 6-1 Introduction 6-2 Fundamental postulates of magnetostatics

in free space 6-3 Vector magnetic potential 6-4 Biot-Savart's law and applications 6-5 The magnetic dipole

6-5.1 Scalar magnetic potential 6-6 - Magnetization ;~ntl cquivalcnt currcnt dcnsiiics 6-7 Magnetic field intensity and relative permeability I

6-8 Magnetic circuits 6-9 Behavior of magnetic materials I

6-10 Boundary conditions for magnetostatic fields 1 6-1 1 Inductances and Inductors

1

. 6-12 Magnetic energy . 6-12.1 Magnetic energy in terms of field quantities

/

I CONTENTS xiii 1

1 . I 8-2.1 Transverse electromagnetic waves 312

8-2.2 Polarization of plane waves I 314 Plane waves in conducting media 1 317 8-3.1 Low-loss dielectirc I 318 8-3.2 Good conductor 8-3.3 Group velocity Flow of electromagnetic power and the Poynting vector 8-4.1 Instantaneous ana average power densities Normal incidence at a plane conducting boundary Oblique incidence at a plane conducting boundary 8-6.1 Perpendicular polarization 8-6.2 Parallel polarization Normal incidence at a plane dielectric boundary Normal incidence at multiple dielectric interfaces 8-8.1 Wave impedance of total field 8-8.2 Impedance transformation wish- multiple dielectrics Oblique incidence at a plane dielectric boundary 8-9.1 Total reflection 8-9.2 Perpendicular polarization 8-9.3 Parallel polarization Review questions Problems

9 Theay and Applications of Mnsmission Lines

Introduction Transverse electromagnetic wave along a parallel-plate transmission line 9-2.1 Lossy parallel-plate transmission lines

General transmission-line equations 9-3.1 Wave characteristics on an infinite transmission line 9-3.2 Tr:lnsrnission-li11c uurumctcl.~ 9-3.3 Attenuation constant from power relations Wave characteristics on finite transmission'lines 9-4.1 Transmission lines as circuit elements 9-4.2 Lines with resistive termination

. I 1 I '! I

I ? 1

I

' , I xiv CONTENTS 1

. . , ;. ' I * < .

* ' . t + . - i . . i I i

( I 1 ' I ' J 1 '"

9-4.3 Lines with Gbj t iaq terminafian . , 404

9-4.4 ~ransmissioa-link circuits . 407 11 An

: 9-5 The Smith chart 41 1 11 1

9-5.1, Smith-chart ca1culations for losiy lines I 420

11 9 -6 Transmission-line im&d&ce matching 1 - . , . 422

9 4 . 1 Impedance rnatc$ng by quarter: wave transformer , 423 ! 1 9-6.2 Single-stub matching 426 11 9-6.3 Double-stub matching 43 1 Review questions 1 . 435 11 Problems , , 437

F

(

1 0 Waveguides and Cavity R8so"ators ,

Introduction General wave behaviors along uniform ' guiding structures , - 10-2.1 Transverse electromagnetic waves 10-2.2 Transverse magnetic waves 10-2.3 Transverse e1ecp;ic waves ,

Parallel-plate waveg$de* t

10-3.1 TM waves petween yarahlel plates 10-3.2 TE waves between parallel plates 10-3.3 Attenuatioq in.~~rallel-plate ' waveguides . t

Rectangular waveguides 10-4.1 TM waves in rectangular wqvebides 10-4.2 TE waves in recrangular waveguides 10-4.3 ~ t tenui i t i$n)rk tan~ular waveguides Dielectric waveguide6 : ! 10-5.1 TM wayes lo^& a dielectric slab 10-5.2 TE waves ;)ong ii dielectric slab '

> , Cavity resonators \

10-6. ! TM,,,, moqes I I , I

10-6.2 TE,,, modts 10-6.3 Quality factor &(cavity resonatbr Review questions Problems

... , I . . . , . . , .

. , , . . , . . , . '

. . . , . . , , , CONTENTS xv . ' ; - ,$ .. ' , ' - 3 . . '. , 1.. . . . . , . . . . . , . < , . , . . . . , , . . . : h - , . ~ z l , ~ . . , , . , ~ . ., -. ,

J - , * . ,, " , - ... ;. --.. .?;,; ... i ; ..: ;>"'.:.',' '

: . ., .. . , ', . ., .. . ., . I , . ;,; , ,:v ,,,, :;. f +.< . ' ,

i . . . . .

: . ,, ' , !!., ., 11 '.. , .., , . i , ~ ' " . ,

Antekms and Radiating Systems. " , . , ,. . . .

, . I S

Introduction ~ ' ~ 1 . , { 500

Radiation fields of elemental dipoles 502 11 -2.1 The elemental electric dipole 11 -2.2 The elemental magnetic dipole Antenna patiefh5 and at. tenna parameters * . Thi;. linear antennas 11 -4.1 The halt-wave aipde Antenna arrays 11 -5.1 Two-element arrays 1 1-5.2 General uniform linear arrays Receiving antennas 11 -6.1 Internal impedance and directional pattern 11 -6.2 Effective area Some other antenna types 11 -7.1 Traveling-wave antenna '-\ 11-7.2 Yagi-Uda antenna 11 -7.3 Broadband antennas Apeiture Radiators References Review questions ProbIems

Amendix A Symbols and Units

A-1 Fundamental SI (rationalized MKSA) units A-2 Derived quantities A-3 Multiples and s~bmultiples of units

Appendix B Some Useful Material Constants

B-1 Constants of free space B-2 Physical constants.of electron and proton

B-3 Relative permittivities (dielectric constants) \ ..I I

I . ,. B-Q ~onductivities , I ! i

t .., . . B-5 Relative permeabilities : ,

Back Endpapers .

Left: . , Gradient, divergence. cprl, and Laplacian opetations Right: Cylindrical coordinates Spherical coordinates .

The many books on introductory electromagnetics can be roughly divided into two main groups. The first group takes the traditional development: starting with the experimental laws, generalizing them in steps, and finally synthesizing them in the form of Maxwell's equations. This is an inductive approach. The second group takes the axiomatic development: starting with Maxwell's equations, identifying each with the appropriate experimental law, and specializing the general equations to static and time-varying situations for analysis. This is a deductive approach. A few books begin with a treatment of the special theory ofrelativity and develop all of electromagnetic theory from Coulomb's law of force; but this approach requires the discussion and understanding of the special theory of relativity first and is perhaps best suited for a course at an advanced level.

Proponents of the traditional development argue that it is the way electromagnetic theory was unraveled historically (from special experimental laws to Maxwell's equations), and that it is easier for the students to follow than the other methods. I feel, however, that the way a body of knowledge was unraveled is not necessarily the best way to teach the spbject to students. The topics tend to be fragmented and cannot take full advantage of the conciseness of vector calculus. Students are puzzled at, and often form a mental block to, the subsequent introduction of gradient, divergence. and curl operations. As ;I proccss for formu1;lting :in clcctroni:~gnctic model, this approach lacks co,hesivoness and elegance.

The axiomatic development usually begins with the set of four Maxwell's equations, either in differential or in integral form, as fundamental postulates. These are equations of considerable complexity and are difficult to master. They are likely to cause consternation and resistance in students who are hit with all of them at the beginning of a book. Alert students will wonder about the meaning of the field vectors and about the necessity and sufficiency of these general equations. At the initial stage students tend to be confused about the concepts of the electromagnetic model, and they are not yet comfortable with the associated mathematical manip- ulations. In any case, the general Maxwell's equations are soon simplified to apply to static fields, which allaw the consideration of electrostatic fields and magnetostatic fields separately. Why then should the entire set of four Maxwell's equations be introduced a t the outset?

'

I vi PREFAC'

It may be argued tfiat'Coulomb's law, i h o u ~ h based on experimental evidence, is in fact also a~postulate.' Cbnsider the tw6 stipulations of Coulomb's law: that the charged'bodies are very spa11 comparec/ with thcir distance of separation, and that the force between the qhafged bodies is isv'crscly proportional to t11c sclu;~rc of their distance. he question a~ises regarding the first stipulation: How small must the charged bodies be in ardqr to be considered "very small" compared with their distance? I,, practice the,charged bodies cannot .be of vanrs' ag sizes (ideal p o h t charges), and there is dificplty in determinirig the?'true" distance between two bodies of finite dimensions. Fgr given body sizes the relative accuracy in distance measurements is better when {he separation is largdr. However, practical considerations, (weakness o f force, existence of extraneous charged bodies, etc.) restrict the usable distance of separation in the laboratory, and experimental inaccuracies cannot be entirely avoided. This {ends to a more importa~if question concerning the invcrse- square relation of thc second stipulnlion. Even if thc clwrgcd bodies wcrc of vanishing . sizes, experiment?! measurements could not be of an infinite accuracy no matter how skillful and careful an experimentor was. ko+ then was it possible for Coulomb to know that the force,wns exactly inversely pr'pportional to thc syuure (not the 2.000001th or the 1.995j999th power) of the distance of separation? This question cannot be answered from an experimental1 viewQoint because it is not likely that during Coulomb'$ time experiments could fiave been accurate to the seventh place. We must therefore conclud'e.that Coulombis lawsis itself a postulate and that it is a law of nature discovered and assumed on the basis of his experiments of a limited accuracy (see Section 3-2). '

This book builds tpe dcctromagnetic inode] using ap axiomatic approach in steps: first for static elecrric fields (Chapter 3), then. for siatic magnetic fields (Chapter 6), and finally for timemrying .fields leading to ~ a x k e f l ' s equations (Chapter 7). The mathematical basis:for each step is Halmhohr's theorem, which states that a vector field is determined to within an additive cbnst$bt if both its divergence and its curl are speciff?;d evfrywhere. Thus, for the::'development of the electrostatic model in free space.. it is 'pnly necessary to define n singb vector (namely. the electric liclcl iulcuaily E) by spccil'~iug its clivcrg~.rde and its cur! as postulates. All other relations in electrostati~s for free space, ihchding Coul~mb's law and Gauss's law, can be derivb- !hq two rather simple postulates. Relations in material media can be devclopedihydugh the concept of eguival&nt charge distributions of polarized dielectrics. r , f

Similarly, for fhe magphstatic model .in .fr& space it is necessary to define only a single magnetic fluyidensity vector B by .specifying its divergence and its curl as postulates all other formulas can be derived frpm these two postulates. E, Relations in rnater!al medip ban be developed t$ough.fhe concept of equivalent current densities. Of coyrsp' the validity o f the postulates lies in their ability to yield results that dpnfoqb pith experimentil evidince.

For time-varying fields, 'the electric andYmagliftic field intensities are coupled. The curl E postula& for)hy electrostatic Godel rmst be modified to conform with Faraday's law. In addition, the curl B postdate the magnetostatic model must also be modified in'prder: to be consistent with of continuity. We have,

= _ > L

* 2 '

8 : I \

. as-.

PREFACE vii

bnental evidence, bb's law: that the hration, and that je square of their v small must the :d with their dis- izes (ideal point !ween two bodies ;istancc measurc- 11 considerations :strict the usable racies cannot be ling the inverse- vere of vanishing m c y no matter ble for Coulomb syuure (not the '! T . n p e s t i o n ; not . ~ ~ e l y that 2 > .* '\ placu. ie and that it is :nts of a'lirnited

~ t c approach in ' fields (Chapter ns (Chapter 7). :h states that a divergence and le electrostatic rly, the electric ates. All other v and Gauss's )us in material lihbutions of

isary to define

rgenwd its wo t .tulates. o i -dent

hei?,dity to

s are coupled. conform with ;c model must l i ty . We have,

then, the. four Maxwell's equations that constitute the electromagnetic model. ,

I believe that this gradual development of the electromagnetic model based on Helmholtz's theorem is novel, systematic, and more easily accepted by students.

In the presentation of the material, I strive for lucidity and unity, and for smooth and logical flow of ideas. Many worked-out examples (a total of 135 in the book) are included to emphasize fundamental concepts and to illustra~e methods for solving typical problems. Review questions appear at the end of each chapter to test the students' retention 2nd ur.dex;anding of the esse~tial material in the chapter. Thc problcms in c;~ch chap~cr arc tlcsigncd rcinforcc >,;,!cnts1 comprchcnsion of the interrelationships b-tween the difTerent quantities in the formulas, and to extend their abilitywf appljling the formulas to solve practical problems. I do not believe in simple-minded drill-type problems that accomplish little more than sn exercise on a calculator.

The subjects covered, besides the fundamentals of electromagnetic fields, include theory and applications of transmission lines, waveguides and resonators, and antennas and radiating systems. The fundamental concepts and the governing theory of electromagnetism do not change with the introduction of new eiectromaz- netic devices. Ample reasons and incentives for learning the fundamental principles of electromagnetics are given in Section 1-1.. I hope that the contents of this book strcngthened by the novel approach, will pr'oiide students with a secure and sufficient background for understanding and analyzing basic electromagneiic phenomena as well as prepare them for more advanced subjects in electromagnetic theory.

There is enough material in this book for a two-semester sequence of courses. Chapters 1 through 7 contain the material on fields, and Chapters 8 through 11 on waves and applications. In schools where there is only a one-semester course on electromagnetics, Chapters 1 through 7, plus the first four sections of Chapter 8 would provide a good foundation on fields and an introduction to waves in unbounded media. The remaining material could serve as a useful reference book on applications or as a textbook for a follow-up elective course. If one is pressed for time, some material, such as Example 2-2 in Section 2-2, Subsection 3-11.2 on electrostatic forces, Subsection 6-5.1 on scalar magnetic potential, Section 6-5 on magnctic circuits, and Subscctions 6-13.1 and 6-13.2 on magnetic forces and torqucs, may be omittcd. Schools on a quarter system could adjust the material to be covered in accordance with the total number of hours assigned to the subject of electromagnetics. '

The book in its manuscript form was class-tested several times in my classes on electromr\gnetics at Syracuse University. I would like to thank all of the students in those classes who gave me feedback on the covered material. I would also like to thank all the reviewers of the manuscript who offered encouragement and valuable suggestions. Special thanks are due Mr. Chang-hong Liang and Mr. Bai-lin Ma for their help in providing solutions to some of the problems.

Syracuse, New York January 1983

\

! 1-1 INTRODUCTION

Stated in a simple fashion, electromugnetics is the study of the effects of electric charges at rest and iq motion. From elementary physics we know there are two kinds of' charges: posithe ahd negative. Both positive and negative charges are sources of an electric field M b ~ i n ~ c h a r g e s produce a current, which gives rise to a magnetlc field. Here we tentatively speak of electric field and magnetic ficld in a general way; more definitive meanings+will bc attached to these terms later. A j e l d is a spatial distribution of a q u a d y , which may or may not b;e, s function, of time. A time-varying electric field is accompanied by a magnetic field, and vice versa. In other words, time-varying electric and magn4iic fields are coupled, r$suiting i i rin electromagnetlc field. Under certain conditibns, time-dependent electromagnetic fields produce waves that radiate from the sotlfce. The concept of fields and waves is essential in the explanation of action at a distance. In this book, Field mii Wave Electromugnetics, we study the principles a d applications of the laws of electromagnetism that govern electromagnetic dhenomepa.

Eleciromagnetks is of fundamental importance to physicists and electr~cal engineers. ElEctromagnetic theory is indispensable in the understanding of the principle of atdm smashers, cathode-ray oscillosco~s, radar, satellite communication, television reception, remote sensing, radio astronomy, microwave devices. optical fiber communication, instrument-landing systems, electromechanical energy conversion, and s? on. Circuit concepts represent a restricted version, a special case, of electromagni?tic caflcepts. As we shall see in chap& 7, when the source frequency IS

very low so that th& dimensions of a conducting ntfwork are much smaller than the wavelenglh, 3 e have a quasi-static situation, which simplifies an electromagnetic problem *circuit problem. However, we hasten to add that circuit theory is itself a highly deveroped, sophisticated discipline. It appHls to a different class of electrical engineering p r~b lebs , rtnd it is certainly important in its own right.

Two sitbations:illustrate the inadequacy of circuit-theory concepts and the need - * - . of electroma~netic-field concepts. Figure !-I depicts a monopole antenna of the type we sce on a wdkie4alkic. OH tnuwmit, thc sotlrcc at thc basc Cccds the antelm1 wlth P mcssagc-carrying currcot d an appropriate carrier frequency. From a circuit-theory

2 THE ELECTROMAGNETIC MODEL / 1

A monopole antenna. Fig. 1-2 An electromagnetic problem.

point ofview, the source feeds into an open circuit because b b e r tip of the antenna is not connected to anything physically; hence no current would flow and nothing would happen. This viewpoint, of course, cannot explain why communication can be established between walkie-talkies at a distance. Electromagnetic concepts must be used. We shall see in Chapter I1 that when the length of the antenna is an appreciable part of the carrier wavelengtht. a nonmiform current will flow :dong thc ol7cn-ended alltellna. This current radiates a tin~c-v;lryiog electrornil~netic field in space, which can induce current in another antenna at a distance.

In Fig. 1-2 we show a situation where an electromagnetic wave is incident from the left on a large conducting wall containing a small hole (aperture). Electromagnetic fields will exist on the right side of the wall at points, such as P in the figure, that arc not necessarily directly behind the aperture. Circuit theory is obviously inadequate here for the determination (or even the explanation of the existence) of the field at P. The situation in Fig. 1-2, however, represents a problem of practical importance as its solution is relevant in evqluating the shielding effectiveness of the conducting wall.

Generally speaking, circuit theory deals with lumped-parameter systems- circuits consisting hf components characterized by lumped parameters such ar rcsisl:~~~ccs, i~~ductancbs, :111d ~ ~ a c i l s ~ ~ c e s . V O I I P ~ C S and currents are the main system variables. For DC circuits, the system variables are constants and the governing equations are algebraic equations. The system variables in AC circuits are time-dependent; they are scalar quantities and are independent of space coordinates. The governing equations are ordinary differential equations. On the other hand, most electromagnetic variables arefunctions of time as well as of space coordinates. Many are vectors with both a magnitude and a direction, and their representation and manipulation require a knowledge of venor algebra and vector calculus. Even in static cases, the governing equations are, in general, partial differential equations. It

' The product of the w&elength and the frequency of an AC source is the vdocity of wave propagation. i

. . I

, . .

. A . <. .-

: r NETlC MODEL . 3 ' , k : , ? '

': :

the antenna wd 110tlling 11ion can be Pbnt bc appr- '-ible O ~ I C I L . .rlcd IXJGC, w l i i d l

udent from tomagnctic ire, that are inadequate e field at P. portance as ucting wall. systems -

:ts such as : the main ~d the gov- circuits are oordinates. hand, most a t e m a r l y .tation and 1 s . in luations. It

f $

is essential that I& ar<eqiiipped to h$jdldvector quantities and variables that i re bath time- ahd &ace-dependent. The hnhamentals of vector algebra and vect6r calculus wil1:be dkveloFed in chapte{ 1. Techniques for solving partial differentip1 equations are heededin Pealing with certain types of electromagnetic problems. ~ h e k techniques will b&diicyssed in Chaptir 4. t h e importa'nce of acquiring a fac~lity in the use of these mathematical toolr'in the study of electromagnetics cannot be

1 overemphasized. ,, , 3 y "

1-3 THE ELECTROMA~$NETIC' MODEL '

I 4 ' . . r '7 There are two adproaches in the deve!c?pn$m 31 i~ scientific subject: the inductive approach and theltieductive approach. .Usink the inductwe approach, one follows the historical develophent of the subject, starthg w ~ t h the observat~ons of some s~mple experiments and i ~fcrring from thcm luws and thcorcms. I t 15 :I process of rc.l\onlng from particular p b enomena to generdl p r ~ d c i ~ ~ e s . The deduct~ve approach, on the other hand, posttnlntes~a few t i~~~dn~~rental : re lnt ions for a n ~dc:~locd rnodel The pohtulalccl rclatid,ils J I G asiw\s. 1'1.0111) \~llicI~ lurlic111~11 I , I W ,111d [ I I ~ O I U I ~ ~ L , I ~ I bc derived. The vd~dity o l ' h ~llodcl .mi the a ~ h s is ver~fed by t h r abdity to prcd~ct consequences thatb chec k with expermental observations. In t h ~ s book we prefer to use thc dcductivc nr :rxlonmllc apprcxtch b c p ~ ~ s c i t is morc clcgant and enables the clcvclolmcl~l oI'lI1~ \11I,jpcl o I 'u lcc l~oin i i~~~cl~c \ In an.ortlcrly w;~y

'lllc idcalitcd modcl we adopt lor slu&~ng :I \cicnLllic suI>jcct must rcl;~tc LO

real-world situahohs and be able to explain physlca~ phenomena; otherw~se, we would be engaged in merltal exercises for no purposc, For example, n thcoret~cal modcl could be bu~lt, from which one might obtain many mathematical relat~ons, but, ~f these relations disagree with observed results, the qbdel is of no use. The mathematlcs may be correct, Btd the hnderlying assumptions o!tfi~model may be wrong or the lmplled approximatiohs d a y not be justified.

Three esskntia) steps are ~nvolved ,in bdildin$ a theory on an idealized model. First, some basic duantities germane ta the iubject of study are defined. Second. the rules of operatiod:(the mathematics) of these quantities are specified. Thud, some fundamental relatqns ore postulated. These postulates or laws are invariablj based on numerous exp&mental observatiops acquired under controlled cond~tlons and synthesized by brgliant hinds. A famiJiar example is the arcult theory built on a circuit model bf ideal sources and pure resistbnces, inductances, and oapacltances. In this case the-basic quantities are voltages (V), currehts (I), resistances (R), inductances (L), and'mpct~itadkes (C); the rules of operatiods are those of algebra, ordmary differential equations, qnd Laplace transfordation; and the fundamental postulates are Kirchhoff's voltage apd current laws. Many relations and formulas can be derived lrom this basiwlly fatho< simplc n ~ o d c l , ~ q ~ ~ d the r d p o o ~ e s of very clilborate neworhs can be determined.,The validity and valueof tli'e model have been amply demonstrated.

In a like manner, qn, electronlagnetic theory can be built on a suitably chosen electrornagnctic madel. lo this section we shall take thc first step of defining the basic

4 THE ELECTROMAGNETIC MODEL I 1 ,'

quantities of electromagnetics. The second step, the rules of operation, encompasses ' vector algebra; vector calculus, and partial differential equations. The fundamentals of vector algebra and vector calculus will be discussed in Chapter 2 (Vector Analysis),

where C is the abbreviation of the unit of charge, c o ~ l o m b . ~ It is named after the French physicist Charles A. de Coulomb, who formulated Coulomb's law in 1785, (Coulomb's law will be discussed in Chapter 3.) A coulomb is a very large unit for .electric charge; it takes 1/(1.60 x 10-19) or 6.25 billion electrons to make up - 1 C. In fact, two 1-C charges 1 m apart will exert a force of approximately 1 million tons on each other. Some other physical constants for the electron are listed in Appendix B-2.

The principle of conscrvutic~n OJ electric churlye, like thc principlc of conservation of momentum, is a fundamental postulate or law of physics. It states that electric charge is conserved; that is, it can neither be created nor be destroyed. Electric charges can move from one place to another and can be redistributed under the influence of an electromagnetic field; but the algebraic'sum of the positive and negative charges in a closed (isolated) system remains unchanged. he principle of conser~ution of electric charge must be sutisjied at all times und under uny circum.stunccs. It is represented mathematically by the equation of continuity, which we will discuss in Section 5-4. Any formulation or solution of an electromagnetiqproblem that violates the principle of conservation of electric charge .must be incorrect. We recall that the Kirchhoff's current law in circuit theory, which maintains that the sum of all the currents leaving a junction must equal the sum of all the currentscntering the junction, is an assertion

and thc techniques for solving partial clilTcrenti;~l c q ~ ~ ~ t i o i i s will hc inrroc111cc.d wl~c~i thcsc equalions wise later in thc bod,. 'l'lic //rid slcp, l l~c I'imlu~iic~ilal pu,\tulntcs, \till

be presented in three sbbst+s in Clinpws 3, (1, and 7 as w e dcal with, rcspcctivoly, static electric fields, steady magnetic fields, and electromagnetic fields.

The quantities in our eleclxoi~~ngnctic niodcl ciln bc divided roughly rftd two catcgorics: source 'r"rL"lle1d qu:uilltics. I'hc source of an electronlagnctic lield is invariably electric charges at rest or in motion. However, an electromagnetic field may cause a redistribution of charges which will, in turn, change the field; hence, the separation between the cause and the effect is not always so distinct.

We use the symbol q (sometimes Q) to denote electric charge. Electric charge is a fundamental property of matter and it exists only in positive or negative integral nit~ltiplcs of rllc cliatgc on an clcctroll. c.+ -. --

-.

' In 1962 Murray Gell-Mann hypothesized qtiarkJ as the basic building blocks of matter. Quarks were predicted to carry a fraction of the charge, e, of an electron; but, to date, their existence has not been verified expericnentally.

The system of units will be discussed in Section 1-3.

e = - 1.60 x 10-l9 (C), (1-1)

spectively;

, into two - tic field ':

c field may henye, the

I

charge is a ve integral

(1 -1)

r) XI af' the I W in . , d5 . gc u n ~ t I'ur

2 up - 1 C. nillion tons 1.4ppendix

~nscrv;ilion hat clcaric tric charges i_nfiuence of tive bharges t i 0 1 1 of c4cc- rcprcscntcd ;corion 5-4. he principle Kirchhoff's ems leaving an n t i o n

tompasses AamentBls Analysis),

1 . - ! [;"' ! '

Iced when ulates, will ,

I ' t - i

.. Q~iIrhs were i not bcen veri-

current law is th; 4

does not exist at a an atomic scale are unimportant

of charges. In conA

results. (The same is defira'i f

as follows: F

'C I 'Aq p = lim - , (C/m3),

A V O AV where Aq is the ambunt of charge in a very smbll volume Av. How small should Ao be? It should be small enough to represent an acctlrate variation of p, but large enough ta contain a very largk nudiber of discrete chnr$es. For exalnple, an elemental cube lbrth

I sides as small as 1 micron (lo-' m or 1 pn1) has a volume of 10- l 8 m3, which wrll st111 contain about 10' (100 billion) atoms. A imoothed-out function of space coordinates. p, defined with sudh a small Aa is expecied to yield accurate macroscopic results for nearly all p c t i c a l pr<oscs.

I ' In somc physical situations, an amount oT chargc Aq may be ident~fied wlth an

element ofsurface As or an element ofline h/..tn such cases, it will be more appropriate to define a surface charge density, p,, or,a line charge density, p,:

* Except for certain bpec:al, situations, charge densities vary from point to point; hence p, ps, and p, are, ih gereral, point functions of space coordinates.

Current is the rate of change of charge dith rcspcct to time; that IS,

where I itselfmay dk time-dependent. The unit of current is coulomb per second (CIS), which isihqsanie hs ampere (A). A curretlt must flow through a finite area (a coni ducting wire of a fitite ctoss section, for instafic6); hence it is not a point functlon. Iti electromagneti~s~yk define a vector poipt function uol~mw cur.r.cnr density (or simply, .current density) J, F h i ~ h measures the amouflt of current flowing through a unit area normal to the direc!ion df current flow. The bold-faced J is a vector whose magnitude is the current per uhit arei ( ~ / r n ~ ) and whose direction is the direction of current flow, We shall elnhordteon tbe.relntion between I and .I in Chapter 5. For very good

4

6 THE ELECTROMAGNETIC MODEL I 1 .4

,.I " ,< L I . 4

I

conductors, high-frequency alternating currents are confined in the surface layer, instead of flowing throughout the interior of the conductor. In such cases there is a need to define a surface current density J , , which is the current per unit width on the conductor surface normal to the direction of current flow and has the unit of ampere per meter (A/m).

I-. . There ark: fo r fundamental vector field quantities in e l ec t romgnc t i~~ , electric /iL4/ i l~(ol~si~jl I(, d ~ ~ l r i i - / / ~ ~ , ~ hwsi/y (or d w f r i y ( /~ ,Y/ ) / ( Ic~~III~~II / ) D, I I W ~ I I I P / ~ / / I I Y (lwsi! Y B. and ~ ~ w g ~ ~ ~ ( i c ~ i e l d i~~(<wsit,v 11. *IIIc ~ l c l ' t ~ ~ i ~ i o ~ ~ : I I I ~ physiul S ~ ~ . I I ~ ~ . * I I I I C C 01' I h c quantities will bc cspl:~il~cd I'ully wllcn tllcy arc introrlilccd hlcr in tllc book. /\t this time, we want only to establish the following. Electric field intensity E is the only vector needed in discussing electrostatics (effects of stationary electric charges) in free space, and is defined as the eledtric force on a unit test charge. Electric displacement vector D is useful in the study of electric field in material media, as we shall see in Chapter 3. Similarly, magnetic flux density B is the only vector needed in discussing magnetostatics (effects of steady electric currents) in free space, and is related to the magnetic force acting on a charge moving with a given ieloctty. The magnetic field intensity vector H is useful in the study of magnetic field in material media. The definition and significance of B and H will be discussed in Chapter 6.

The four fundamental electromagnetic field quantities, together with their units, are tabulated in Table 1- 1. In Table 1- 1, V/m is volt per meter, and T stands for tesla or volt-second per square meter. When there is np time variation (as in static, steady,

Table 1-1 Electromagnetic Field Quantities

Symbols and Units for Field Quantities

Electric

or stationary cases), the electric field cpantities' E and D and the magnetic field quantities B and H form two separate vector pairs. In time-dependent cases, however, electric and magnetic field quantities arc coupled; that is, time-varying E and D will give rise to B and H. and vice versa. All four quantities are point functions; they are defined at every point in space and, in general, are functions of space coordinates. Material (or medium) properties determine the relations between E and D and between B and H. These relations are called the constitutive relations of a medium and will be examined later.

Magnetic

Field Quantity

Electric field intensity

Electric flux density . (Electric displacement)

Magnetic flux density

Magnetic field intensity

Symbol

E

D

Unit

v /m

C/m2

U

H

T

A/m

face layer, in- here is a need h on the con- 2f ampere per

etics: electric ic. Pus density anre uf these book. At this E IS tne only ~ x g e s ) in f~ L,. displacement e shall see in in discussing elated to the ugnetic field

11 t lv ' llnits, mds A,, tesla itatic, steady,

Unit E

E ail$ " will Ins; t l . - j are coordinates. and 0 and medium and

J I. 1-3 h SI U?!:T,E AND UNIVERSAL CONSTANTS 7 I

X . L

' ' I ' L

The principal bbjective of studying.el~ctiomagnetism is to understand the inter; action between charges land currents a t a. distance based on the electromagnetic model. Fields. and: waves (time- and s&ehependent fields) are basic conceptual quantities ofthis rrlodel. Fundamental postulhes will relate E, D, B, H, and the source quantities; and dkrived:reiations will. leid to the explanation and prediction of electromagnetip

1-3 SI UNITS AND* :, UN~VERSAL CONSTANT$ I . .

A me&, , ;n,c,~i i)ikij pi~lysicai quantity mu t be expressed as a number followed by a unit. Thus, we mdy talk about a length of th 1 ee meters, a mass of two kilograms, and a time-period of teh seconds. To be useful, u&t system should be based on some fundamental units of convenient (practical) sizes. In mechanics all quant~ties can be expressed in terms bf three basic units (for length, mass and time). In eiectromagnetics work ;I fowl1 basic unit (for cusrcnl) is ~~cctlcd. Thc SI (Irrtc~r~rrc~rror~lrl S!vcnl (4 L ~ r ~ l r \ or Le Sy~1L;nre lder.~rafto,de d'U11itbs) is an MKSA S ~ J ~ ~ I ) I budt from the four fundamental units listed in Table 1-2.. All other units used In electromagnetlcs, including t h d e appearing in Table 1-1, are derived unlts expressible In terms of m, kg, s, and A. For example, the unit for charge, coulomb (C) 1s ampere-second (A.s); the unit for electric field intensity (V/m) is kg:&m/A.ss; and the unlt for magnetlc flux density, tesla (T), is kg/A.s2. More complete'tables of the units for varlous quantities are given in Appendix A.

In our electromagnetic model there are three universal constants, in addit~on to the field quantitie9,listed in Table 1-1. They relate to the properties of the free space (vacuum). They arc! as follows: velocity of electromagnetic wave (including light) in free space, c; p~rmittivity of frce space, E,; and permeability of free space, p,. Many experiments have been performed for precise measurement of the velocity of light, to many dlcimal places. For our purpose, it id sufficient to remember that

Table 1-2 Fundamental St Units --l

Quantily Unil Abbrcvintion

Length I Deter 1 m kilogrtlm

Tim'e s

~urreni 1 ampere 1 A

8 THE ELECTROMAGNETIC MODEL I 1 i - \

The other two constants, E, and p,, pertain to electric and magnetic phenomena respectively: E, is the proportionality constant between the electric flux density D and the electric field intensity E in free space, such.that

p,, is the proportionality constant between the magnetic field intensity H 2nd the magnetic flux density B in frce spacc, such tliat

The values of E, and pO are determined by the choice of the unit system, and they are not independent. In the S I system (rationalizedt MKSA-sys~m), which is almost universally adopted for electromagnetics work, the permeability of free space is chosen to be

REV ' -

I po = 4n x lo-' (Him), I (1-9)

where H/m stands for henry per meter. With the values of c and po fixed in Eqs. (1 -6) and (1-9), the value of the permittivity of free space is then derived from the following rclntionships:

where F/m is the abbreviation for farad per meter. The three universal constants and their values are summarized in Table 1-3.

Now that we have defined the basic quantities and the universal constants of the electromagnetic model, we can dcvclop thc various subjccts in clectromagnctics. But,

' This system of units is said to be rationalized because the factor 471 does not appear in the Maxwell's equations (the fundamental postulates of electromagnetism). This factor, however, will appear in many derlved relations. In the unrationalized MKSA system, p, would be lo-' (H/m), and the factor 4n would appear in the Maxwell's equations.

Permittivity &ree space , :y H and the .. $ - i '-

0 .,-a

B Y 3, . . (1-8) ' ! before we dqthat,~be mustbe~eqequipped witH the appropriate mathematical tools. In

the followihg chapter, we discuss the basic rtdes of operation for vector algebra and vector calculus. i -

- 1

.. ' > i t. d : . REVIEW QUESTIONS 9 ,

I ., 1; q . "

a- ' 1; ..w jS able 1-3 y?iversal Constints in SI Units 1 , c phenimena

1,' : k . A

ux density D Universal &dnstantst ' l i

em, and they ich is almost

1

h e space is REVIEW QUESTIONS

r; R.l-1 What is electromagnetics? '(1 -9)

[ '

! .

Velocity of light in free space

F (1 -7) Permeability of free bbace i

R.1-2 Describqtwo phenomena or situations, otnkr than those depicted in Figs. 1-1 and 1-2, that cannot be adeqlftltely explained by circuit thedry.

in &s. (1 -6) he following R.1-3 What are the three essential steps in building an idealized model for the study of a scientific

subject? . ,

dynibof II

F L

- Po

Value

3 x lo8

471 x lo-'

R.l-4 What are the four fu?drlmental SI units in efectromagnetics? 8

(1-10) R.1-5 What are the four fundamental field quanthies in the electromagnetic model? What are / their units?

Unit

m/s

H/m

I

R.1-6 What are. the hhree ukversal constants in tHe electromagnetic model, and what are their relations?

I R.l-7 What are the source quantities in the electromagnetic model? 1

Instants and

(? Stall., af the

the ax well's ppmr in many ICIOI I n wwld

2 / Vector Analysis

2-1 INTRODUCTION

As we noted in Chapter 1, some of the quantities in clectromngnetics (such as charge, currcnt, cncrgy) arc scn1:~rs: and sotnc oll~crs (such ;is clcclric and magnetic licld intensities) are vectors. Both scalars and vectors can be functib'iis-f time and position. At a given time and position, a scalar is completely specified by its magnitude (positive or negative, together with its unit). Thus, we can specify, for instance, a charge of - 1 pC at a certain location at t = 0. The specification of a vector at a given location and time, ,on the other hand, requires both a magnitude and a direction. How do'ive specify the direction of a vector? In a three-dimensional space three numbers are needed, and these numbers depend on the choice of a coordinate system. Conversion of a given vector from one coordinate system to another will change these numbers. However, physical laws and theorems relating various scalar and vector quantities certainly must hold irrespective of the coordinate system. The general expressions of the laws of electromagnetism, therefore, do not require the specification of a coordinate system. A particular coordinate systctn is choscn only whcn a probl'em of a given geometry is to be analyzed. For example, if we are to determine the magnetic field at the center of a current-carrying wire loop, it is more convenient to use rectangular coordinates if the joop is rectangular, whereas polar coordinates (two- dimensional) will beemore appropriate if the loop is circular in shape. The basic electromagnetic relation governing the solution of such a problem is the same for both geometries.

Three main topics will be dealt with in this chapter on vector analysis:

1. Vector algebra-addition, subtraction, and multiplication of vectors.

2. Orthogonal coordinate systems-Cartesian, cylindrical, and spherical coordinates.

, Ai

3. Vector calculus-differentiation and integration of vectors; line, surface, and volume integrals; "del" operator; gradient, divergence, and curl operations.

Throughout the reit of this book, we will decompbse, combine, differentiate, integrate, and otherwisc manipulate vectors. It is imperutive that one acquire a facility in vector

2-2 ' VE( AND SUE

1 i ' .

i

i

such as charge, magnetlc field c 'ind position. lifudc (positive

: q r g e of I . cation I n . l i 30 wc : numbers are n. Conversion 11f.q~r t l ~ ~ l ~ l ~ ~ ~ r ~ ,

-lor quanlitles expressions of itlon of a co- problem of a the magnetic nt to use rec- dinates (two- pe. The basic the same for

s u d and :rations.

Ire. integrate, lily in vector

I '

, .. ; g h ! ! i "

algebra and Vectof cdculus. In a three-dimensional space a vector relation is, in fan, three scalar rel+t.idns The use of vectoi-analisis techniques in electromagnetics leads to concise and'eidgaht '~orrnulations. A hefieiency in vector analysis in the study Of

, electromagnet& 1; similar to a deficiency .in algebra and calculus in the study of physics; and ift is qbvious that these deficienilies cannot yield fruitful results.

In solving prd~tica~~problems, we plwa$ deal with regions or objects of a given shape, and it is ,necessary to express gqngt.ar'fotipulas in a coordinate system appropriate for the,$yhn geometry. For ek$m$, %$ familiar rectangll!ar (x, y, :) cd- oqkiates arei ib$ouslg awkward to pie f@ brodems involving ; circular cylinddr or a sphere, peca4se ,t@ boundaries c$ a ckcular kylinder and a sphere cannot be describdh by kondtant ;.slues of x, y, and z. ffi this dhapter,we discuss the three most commonly~bs'ed dhhogbnai (perpendicular),coordinate systems and the representation and oneratioh of'vcctors in these systems. Familaritv with these coordinate systems is essential in the solution of electrorhagnetic problems.

Vector ,~2lculhs pertains to the differentiation and integration of vectors. By defining certiin differential operators, wc ,tan express ~ h c basic laws of electro~ magnetism in a cohcisefway that is invariant with the choice of a coordinate system. In this chapter wc introducc thc techniques for evalu:lting dllkrent types of integrals involvinl: Vectors, kind define and discuss the various klnds of differential operators.

. ,

2-2 VECTOR A D D ~ T I O ~ 4

AND SUBTRACTlClN

We know that a vector has a magnitude and a direction. A vector A can be written as

$

where A is the magnitude (and has the unit and dimension) of A, . ' '1

and a, is a dimendionless unit vectort with a unity magnitude having the direction of A. Thus,

The vectmAxan be represented graphically by a directed straight-he segment of a length IAl = .-I with its arrnwhcnd pointing irlllhe c\irection ofa , , a s shown in Fig 2-1, . a

1 wo vcctors are equal if Lhcy have the same hci&iitudc and the same direction, even ..though they may be displaced in space. Sin$ it is difficult to write boldfaced letters by hand, it is a common practice to use an arrow or a bar over a letter (A or A) or ---- -:: ' ..

In some books the unit vector ~ r i the direction of A is arml lid) denoted by b, u,, or i,.

2-3 PRODU 1 W O vectors A B, which are nbt in the same direction nor in opposite direc- such 3s given in Fie. 2-2(.1). dctcrmine :I pl:~ne. Their slim i s i l s , t l l t > r ycrtl>l c hlulll

In the s;lmc p l ; l ~ . C = A + 15 all1 b~ ubt;liled gr:~pllicllly ill ibvo w;lys,

1. BY the parallelogram rule: Tile resultant C is the di;lgonal ved&~tl le Flr;,llclo- I ' Fclm rorllld IJY A lllld 1) dr;lWll kolll lllc S:lll]c poill(, as sllot\r\.ll in Fig :-2(b). I t

2. BY the head-to-tail rule: The head of A connects to the tail of B. Their sum c is uct of the vector drawn from the tail of A to the head of B, and vectors A, B, and C form of twc a triangle, as shown in Fig. 2-2(c).

-B = ( -a , )B.

The operation represented by Eq. (2-6) is illustrated in Fig. 2-3.

Y" Bz A. ." .

A A A (a) TWO vectors. (b) Parallelogram rule. (c) Headia-rai] rule.

Fig. 2-2 Vector addition, C = A + B. k B cos

1 (a) Two vectdrs. I ; (b] Subtract~on qf ., , r ' This distin- I . . vectors, A -:B. , Fig. 2-3 x Vector subtrd~Tion.

rever vectors P . ! . . A .

F . ; ' i i l . ?

5 ' rposite direc- 2-3 PRODUCTS . Q F - ~ X ! ~ O ~ . . . : A

her vector C 1 1 1 ~ u l t i ~ l i c a t i o n of k vector A by a positive scalar i. changes the magnitude of A b i k litncs wilho~lt chllnging its dircclion (k cJn be either grcatcr or less than 1).

I

he parallelo- I kA = a,(kA). (2 - 8) 2-2(b). 1 i It is not safficiek to say "the rn~lti~licatiotl of one vector by another" or .'the prod- c1r '.' C is 1 uct of two vectors" because there are two distihct and very diflerent types of pro,ducts : i i T f o r m of two vectors. The$ arc (1) scalar or ddt probucts, and (2) vector or cross products.

f

! These will bc defined in the foilowtng subsections.

I 1 I 1

:~ulivc laws. 1 2-3.1 Scalar or ~ o t ~roduct

(2-4) !

The scalar or dot produci of two vectors A and B, denoted byA . B, is a scalar, whxh (2-5) t , equals the product bf theimagnitudes 0f.A and B and the cosine of the angle between I

owing way: them. Thus, i i, - (2-6) i (2-9)

k as B, but i 3'

In Eq. (2-9), the syhbol 4 signifies "equal by definition" and O,, 1s the 3mallrr angle between A and B dhd i j less than rr radians (j80c), as indicated in Fig 2-4. The dot product of two vedors (1) is less than or' equhl to the product of their magnitudes; (2) can be either a positive or a negatiYe quihtity, depending on whether the angle between them is smhller or larger than 7t/2 radians (90'); (3) is equal to the product of

t

Fig. 2-4 1llusktfing the dbt product of A and k.

14 VECTOR ANALYSIS 1 2 , \

the magnitude of one vector and the projection of the other vector~upon the first one; and (4) is zero when the vectors are perpendicular to each other. It is evident that

Equation (2-11) enables us to find the magnitude of r vector when the ~~pkessioi l of ?he vector is given in any coordinate system:

The dot prcduct is commutative and distrilx~tivc, -. , l a * , I r ,

I *

Co~lmutative law: A B = U A . (2-12) Distributive law: A . (B + C) = A . B + A . C . (2-13)

The commutative law is obvious from the definition of the dot product in Eq. 12-9), and the proof of Eq. (2-13) is left as an exercise. The associative law does not apply to the dot product, since no more than two vectors can be so multiplied and an expression such as A - B . C is meaningless. - 1.

Example 2-1 Prove the law of cosines for a triangle. .,

Solutioti: The law of cosines is a scalar relationship that expresses the length of a side of a triangle in terms of the lengths of the two other sides and the angle between them. Referring to Fig. 2-5. we find the law of cosines states that

- - - - - C = a + B2 - 2AB cos a .

We prove this by considering the sides as vectors; that is

C = A + B .

Taking the dot product of C with itself, we have, from Eqs. (2-10) and (2-13).

C2 = C . C = ( A + B ) - ( A + B ) - A - A + B - B + 2 A - B = A' + B' + 2AB cos U,,.

\ ,

Fig. 2-5 Illustrating Example 2-1.

Not (1 8C

and

2-3.2 Vec

Hen the c

Can corn

, r , * 1 - ; ,[ : b q

the first one! Note that 8,. is, db debition, the smdlldr Mgle between A and B and is equal t i :nt that , (180" - a); hence, bs @dB = cos (180" +,a) 4 -cos a. Therefore, , > $ , - ,: ? i s 1

(2-10) t 1 C2 = +pr - ; ~ A B Cos a, ' i .

, ' and the law of coslher f o l l o ~ s directly. i / '

- i I \

In Eq. (2-9), I les not apply d and an ex-

i 9

The vector or ~- ;~$uct 4 f two ve+tdrs A and B, denoted by A x B, is a vectdr plafle containing A and Il: its m?mitude is AB sin O,,, whr.re

I).. in ';I.. smclkr -!bglfix betwre:i A m d p, ,;la ~ t s c,irection follows that of t h e 9 2 3 3 of the right hand when tge fingers rotate frontA to B through the angle 8,, (the right- hand rule.)

(2-14) !

This is illustrated ifi Fig. 2-6. Since B sin 8,, is the height of the parallelogram formed by the vectors A and 8, we recognize that the magnitude of A x B, IAB sin B,,I, which is always positive, 16 numerically equal tdthe Brea of the parallelogram.

Using the dcfitlitionrin Eq. (2-14) and following the right-hand rule, we find that -

I

B x A = - ' A x B . (2-15) i

Hence the cross prbduct is nor commutative. w e can see that the cross product obeys the distributive law, 1

Can you show this in: general without resolving the vectors'into rectangulal components?

The vector prdduct is obviously noa associative; that is,

; A x ( B x C ) # ( A x B ) x C .

16 VECTOR ANALYSIS 1 2

The vector representing the triple product on the left side of the expression above is perpendicular to A and lies in the plane formed by B and C, whereas that on the right side is perpendicular to C and lies in the plane formed by A and B. The order in which the two vector products are performed is thcrcforc vital and in no case .\houlil rile parentheses be omitted.

. Product of Three Vect.2~

I

There are two k y d s of e io i:m of l h ~ c scows: nmcly , l l ic a c & s i r i p l t / ~ ~ * o d ~ , r r and t l ~ c tvc~or* t r r p k yrotluci. 1 ' 1 1 ~ scal:ir triple product is mucli the si~nplcr of the two and has the following property:

A . ( B x C ) = B . ( C x A ) = C . ( A x B). (2-18)

Note the cyclic permutation of the order of the three vectors A, B, and C. Of course, \---

A . ( B x C) = - A . ( C x B)

= - C . (B x A). (2-19)

As can be seen from Fig. 2-7, each of the three expressions in Eq. (2-18) has a magnitude equal to'the volume of the parallelepiped formed by the three vectors A, B, and C. The parallelepiped has a base with an area equal to IB. x CI = [BC sin 8,1 and a height equal to \A cos O,l; hence the volume is ]ABC sin 8, cos 8,[.

The vector triple product A x (B x C) can be expanded as the difference of two simple vectors as follows:

Equation (2-20) is known as the "hack-cab'' rule and is a useful vector identity. (Note "RAC-CAR" on the right side'of the cq~~iition!)

I

t ;

i

-- - !

Fig. 2-7 . Illustrating scalar triple product A . (B x C).

. - - . . ..& . . . . . . - ..,; . '." ..,7,.

, , ,. ... . , < ' . , . 6 ' ' .

1 1 , . . I . , I '

, . . . . . .- ...A*.. . , . . . b , , . ' , . . '

. .

2-3 / PRODUCTS OF VECTORS 17

:ssion above is at on the right order in which - use should the ,

- tripir product pler of the two

i C . Of course,

)) ha, ., ,nagni- :tors A, T3, and >in 0 , ) and a

!I'crence of two

identity. (Note

Example 2-2t Prove the back-cab rule of vector triple product.

Solution: In order to prove Eq. (2-20), it is convenient to expand A into two components

where .4,, and A, are, respectively, parallel and perpendicular to the plane containing B and C. Because the vector representing (B x C) is also perpendicular to the plane, the cross pioduct of A, and (B x C' vanishes. Let D = A x (B x C). Since only All is effective her;, ~YI: h;ve

D =.4,1 X (B X C).

Referring to Fig. 2-8, which shows the plane containing B, C, and Al l , we note that D lies in the same plane and is normal to A,,. The magnitude of (B x C) is BC sin (0, - 0,) and that of All x (B x C) is AllBC sin (8, - 0,). Hence,

D = D a, = AllBC sin (8 , - 0,) = (U sin O,)(AllC cos 0,) - (C sin 02)(AIIB cos 0,) = [B(AII C) - C(AII B)] .a,.

Fig. 2-8 Illustrating the back-cab rule of vector triple product.

The expression above does not alone guarantee that the quantity inside the brackets to be D , since the former may contain a vector that is normal to D (parallel to All); that is, D . a, = E . a, does not guarantee E = D . In general, we can write

B(A,~ . cj - c(A~, B) = D + k ~ ~ ~ ,

where k js a scalar quantity. To determine k, we scalar-multiply both sides of the above e q u h n by All and obtain

The back-cab ruk can be verified in a straightforward manner by expanding the vectors in the Cartesian coordinate system (Problem P.2-8). Only those interested in a general proof need to study this example.

I ' r .

. * ' 1 ' i I $(. l' ,'%4 18 VECTOR ANALYSIS / 2 ",*,- :, f - . ,

. . 1 2 , < T

Since All . D = 0, so k = 0 and , , I 1 wh

D=B(A;I I -C)-C(AII iB) , i the the

which proves the back-cab rule jnapmuch as All . C = A . and All B = A B. ? .

Division by a vector is not $ejnt!d, a r d expr?ssions sluch as k/A and B/A are k

I ..- meaningless. qx: (b)

2-4 ORTHOGONAL COORDINATE SYSTEMS ,

We have indicated before that although the laws of'electromagnetism are invariant with coordinate system, solution of practical problert~s requires that the relations derived from thcsc laws he cspscwil in ;I cocrrdi~~.~lc syshm ;~ppropsi:~~c t o lllc geomctry of thc given problems. I:or cxampl~., if wc are to determine the electric field at a certain point in space, we gt least need to describe the posilion of the source and the locatlon of this point in p coordinate system. In a three-dimTnsiona1 space a point can be located as the intersection of three surfaceq. Assume that the three families of surfaces are described by u, = constant, u, = coqstant, and u, = constant, where the u's need not all be lengths. (In the familiar Cartesian pr rectangular coordinate system, u,, u,, and u, correspond to x, y, and z respectively.) When these three surfaces are mutually perpendicular to one another, we have an orthogonal coorriinate system. Nonorthogonal coordinate systems are not used because they complicate problems.

Some surfaces represented byau, = constant (i = 1,2, or 3) in a coordinate system may not be plancs; they may b~ curvkd surhccs. Let a,,, a,,, and a,,, bc the unit vectors in the three coordinate directions. They are callqI the buse vectors. In a general right-handed, orthogonal, curvilinear coordinate system, the base vectors are arranged in such a way that the following relations ace satisfied:

a,, x a,, = a,, , I (2-21a) a,,, >j a,, = a,, I

I (2-21 b)

, aK3:x a,, = a,*. ' (2-21c)

These three equations are not alliqde&4dent, as the spbificaiion ofone automatically implies the other two. We have, qt cqube, I

' - . 1

and

Any vector A can be written as the sum of it$ cornpanents in the three orthogonal . directions, as follows: " \ ! I

1 ' I

11 perfo

2-4 / ORTHOGONAL COORDINATE SYSTEMS 19

where the magnitudes of the three components, A,,, A,,, and A,,, may change with the location of A; that is, they may be functions of u,, u,, and u3. From Eq. (2-24) the magnitude of A is

= A . B . : A = /A/ = (A;$ + A ; ~ + A;~)'". (2-25)

.nd S/A are Example 2-3 Given three vectors A, B, and C, obtain the expressions of (a) A . B,

I (b) A x B, and (c) C (A x B) in the orthogonal curvilinear coordinate system a / ( ~ 1 , U2,"3).

I ,re invariant 1 . Solution: Firs? we write-A, B and C in the orthogonal coordinates (a,, u,, u3):

, he relations 1 A = a,,,Aul f %,Au2 + %,A,,, ricite to the clcctric iicld \

,re invariant I. Solution: Firs? we write-A, B and C in the orthogonal coordinates (a,, u,, u3):

, he relations 1 A = a A + s A + aU3A,,,

f the source I

ma1 space a at the three

L = AuIBul + Au2Bu2 + Au3Bu3, (2-26)

in view of Eqs. (2-22) and (2-23). 1 the?- three I ~ w t r b, A x B = (%,A,,, + ~ , , , A , , , 4- %1/1,,1) x (a,,,B,, + a,,,B,,, + a,,,B,,,) co~npl~cl lw sx %I( A112u143 - A l l , ~ u 2 ~ 4- ~ ~ l , L ( ~ u L ~ ~ , , , - ~ ~ ~ l j , , ~ -I- ,(A,,,LY~~ - Au2Bul)

mate system be the unit ecrors. In a 1

: vectors are f (2-21a)

Equations (2-26) and (2-27) express, respcctivcly, the dot m d cross products of two vectors in orthogonal curvilinear coordinates. They are important and should be remembered. - i

(2-21b) k c) The expression for C (A x B) can be written down immediately by combining !

(2-21c) the results in Eqs. (2-26) and (2-27).

1- Eq. (2-28) can be used to prove Eqs. (2-18) and (2-19) by observing that a

* permutation of the order of the vectors on the left side leads simply to a rearrange- , ment of the rows in the determinant on the right side.

(2- 24) In vector calculus (and in electromagnetics work), we are often required to perform line, surface. and'volume integrals. In each case we need to express the

i

1 20 VECTOR ANALYSIS 1 2 %

I I r I I

I

differential length-change coriesponding to .a differe'ntial ~hange in one of the coordinates. However, some of the coordinatesy say u,. (i = 1, 2, or 3), may not be a length: and a conversion factor is needed to convert a differential change du, into a change in length dt, :

(I/', = It, (Ill,, ( 2 20) S 1 where hi is called a metric coefic(en( a id may itrclf L a fynction of u,, u,, and u,. For example, in the two-dimensional polar coordinat& (u,, h) = (r, 4), a difrcrential change d4 (=du2) in 4 (= 7 l Z ) cor&onds to a differential length-ct~ange d l = r r l$ (h2 = r = u , ) in the a+ (= - :~~ , ) -~ ! i r~~c I io~ l . A dirc~tc,l di l lhc~~ii :~i I C I I ~ I ~ I ~ - I ~ : I I , ~ c i l l : I I I

Ill1 arbitrary direction can be wrilkn i \ s tbevec~or sum uflllc c.oo;pollmt l c t~~ t l l cIlanses:t

dt = a,,, dCI + aU2 dt2 + a,, d& . .

= [ ( h , du!)' + (h2 h,)' + h, J u , ) ' ] ' ~ ~ . (2-32) a

he differential volume dv formed by differential coordinate changes du,, du,, and in directions a.,, a,, and a,, respect~vely is (dll dt2 d!,), qr

Later we will have occasion to express the current or flux flowing through a differential area. In such cases the crdss-sectional area perwndicular to the current or flux flow must be used, and it is convenient to consider the Pifferential area a vector with a direction normal to the surface; ihat is.

ds.= a, ds. . I 1 I I I For instance, if current de;sity J ia ndt pcrpcndiculnr to a diLrcntinl area of a ,nag-

nitude ds, the current, dl, flowing through ds must be the ckponen t of J normal to the area multiplied by the area. Usingthe notation in Eq. (2-34), we can write simply

( ~ r ; = J JS : I . $ , J a,&. (2-35)

In general orthogonal curvilinear coordinates, the 4iff(rentiil area ds, normal to the unit vector a,, is . :. I

,_I ' ds1 = a,,(dd, dt,) ' , J $

4 , ' This e is the symbol of the vector d. 9 , I (

! I . ,I , I . .I , 1

1 1 .

'p the

1.

Thi

2-4.1 Car

A r spec syst

, , . . 2-4 1 ORTHOGONAL COORDINATE SYSTEMS 21 '

me of the co- t 4t r ,Z or may not be a :$ F- I . L f nge dui into a c

1 ,; (2-36) -

u,, and u,. . a differential ge d l2 = rd$ change in an gth changes:'

(2- 30)

ig through a 3 the current area a yector

.ea of a mag- J normal to write simply

n - 3 5 ) ormal to the

Similarly, the differential area normal to unit vectors a,, and a,, are, respectively,

and

ds, = a,,(h,h, du, du,)

I ds, = au,(hlh2 du, du,). I

(2-37)

Many orthogonal coordinate systcrns exist; but we shall only be concerned with the three that are most common and most useful:

I. Cartesian (or rectangular) coordinate^.^ 2. Cylindrical coordinates.

3. Spherical coordinates.

These will be discussed separately in the following subsections.

2-4.1 Cartesian Coordinates

A point P(xl, y, , z , ) in Cartesian coordinates is the intersection of three planes specified by x = x,, y = y,, and z = z , , as shown in Fig. 2-9. It is a right-handed system with base vectors a,, a,, and a, satisfying the following relations:

a, x a, = a,

\ a, x a= = a, a, x a, = a,.

The position vector to the point P ( x , , y , , 2,) is

A v e w in Cartesian coordinates can be written as

-

The term "Cartesian coordinates" is preferred because the term "rectangular coordinates" is custornarlly associated with two-dirnensio'na~geomztry.

I

6 . .

22 VECTOR ANALYSIS 1 2

I

Fig, 2-9 Cartesian coordinates.

- -.--- The dot product of two vectors A and B is, from E& (2-26),

A t B = AxBx + AyBy + A&, (2-42)

and the cross product of A and B i$ from Eq. (2-271,

f

Since x, y, and z are lengihs themselves, all three etric coefficients are unity; ill that is. = h 2 = h, = 1. The expr&ons for the differe tial length, differential area, and differential volume ark - Iroy Eqs. (2-31). (2-36), $-37). (2-38), and (2-33) - respcctivcly, . .

(2-45a) ds, = a, dx dz (2-45b) ds, = a, dx d y ; , (2-45c)

(2-46) - ,

t

1.: Example 2-4 A scalar line integral of a vector field of the type

jp: F . dt'

is of considerable importance in both physics and electromagnetics. (If F is a force, the integral is the work done by the force in moving from P1 to P2 along a specified path; if F is replaced by E, the electric field intensity, then the'integral represents an

I electromotive force.) Assume F = a2;y + aY(3x - y2). Evaluate the scalar line E, t *

integral from P,(5,6) to P2(3, 3) in Fig. 2-10 (a) along the direct path @, PIP2; then I. (b) along path @, P,AP,. P

s are unity; rentid area, .-

ld (2-33) -

Fig. 2-10 ' Paths of integration (Example 2-4).

Solution: -First we must write the dot product F . d t in Cartesian coordinates. Since this is a two-dimensional problem, we have, from Eq. (2-44),

It is important to remember that dt' in Cartesian coordinates is always given by Eq. (2-44) irrespective of the path or the direction of integration. The direction of integration is taken care of by using the proper limits on the integral.

Along direct path a - The equation of the path PIP2 is

This is easily obtained by noting from Fig. 2-10 that the slope of the line P I P 2 is f. Hence y = ($)x is the equation of the dashed line passing through the origin and parallel to PIP,. Since line inte intersects the x-axis at x = + I, its equztion is that of the dashed line shifted one unit in the positive x-direction; it can bs

obtained by replacing x yith (x - I). We have, from E ~ S . (2-47) and (2-48),

Spy F .dP = Spy [ x y dx + (3x - y2) dy ]

Path @ Path t

In the integration with respect to y, the relatioq 3x = 2 y + 3 derived from E q . (2-48) was usede<

b) Along path @ - This path has two straight-line segments:

From P , to A : x = 5 , dx = 0.

From A t o P 2 : y = 3 , d y = 0 .

I;. d t = 3x d x , Hence,

Path

We see here that the value of the line integral hepen+ on the path of integration. In such a case, we say that the vector field F is not conservative.

', . I 2-4.2 Cylindrical Coordinates

' In cylindrical coordinates a point ~ ( r , , 4,. i,) is the intersection of a circular cylindrical surface r = r , , a half-plane cpntaining the z-axis hnd making an angle 4 = 4, with the xz-plane, and a pl ne frallel to the xy-plane at z = 2, . As indicated in 't Fig. 2-11, angle 4 is measured from the x - iuk and the base vcctor a, is tangential to the cylindrical !vrf&e. The following right-hand relations apply: ,

I . d ' " '^' '

I

ri

integration.

rcular cylin- ngle 4 = 4 , indicated in v e r p a , is 1Pk 9 .

( ja) (2-49 b)

(2-49~)

x- 9 = 91 plane ''4 Fig. 2-11 Cylindrical coordmates.

Cylindrical wordinales are important for problsms wilh long line charges or curmnt,, and in places where cylindrical or circular boundaries exist. The two-dimensional polar coordinates arc a special cusc at z = 0.

A vector in cylindrical coordinates is written as

The expressions for the dot and cross products of two vectors in cylindrical coordinates follow from Eqs. (2-26) and (2-27) directly.

Two of the three coordinates, r and r (u, and u,), are themselves lengths; hence h , = h3 = 1. However, q5 is an angle requiring a metric coefficient h, = r to convert d$ to d 4 . The general expression for a differential length in cylindrical coordinates is then, from Eq. (2-31):

(2-51)

The expressions for differential areas and differential volume are

' and

( 1

26 VECTOR ANALYSIS / 2 '

A typical differential volume element at a point (r, 4, 3 resulting from differential changes dr, d 4 , and dq in t& three orthogonal coordinate dir2tions is shown in Fig. 2-12.

A vector given in cylindrical Fodrdinates can be lranlformed into one in Cartesian '

coordinates, and vice versa. Spppose we want to expresi A = a J r + a d r n + a,A, in I

Cartesian coordinates; that iq, w~ want to write A as a,A, i a,A, + a,A, and determine A,, A,, and A,. First ofall, we note that A,, the I-component ofA, is not changed by the transformation from cylindrical to Cartesian coor inates. To find A, we equate 4 - the dot products of both expressions of A with a,, Thus,

i ,

A x =.A . a, , 5 Ara, ax + Ada, A,. i

I The term containing A, disappei)rs here because a, a, = 0. Referring to Fig. 2-13, which shows the relative position's of the base vectors q,, a,, ar, and a+, we see that

' a4.p,=L"s(:+4)=:-sis*. 1

Hence, * i I.

41 = A. cos 4 - Ad sin 4 . :. II

0

. i

------ . !' I

!! Fig. 2-13 Relations between %,,a,, a,. and a*. ji :

. . 1. I I ' I

1 differential is shown in

Fig. 2- 13, we see that

. - 2-4 1 ORTHOGONAL COORDINATE SYSTEMS 27

f

Similarly, to find A,, we take the dot products of both expressions of A with a,:

A, = A a,

= Arar . a, + A,a, . a,.

From Fig. 2-13, we find

a, . a, = cos (: - $) f 4

and

a , . a, = cos 4. It follows that

It is convenient to write the relations between the components oia vector in Cartesian and cylindrical coordinates in a matrix form:

Our problem is now solved except that the cos 4 and sin 4 in Eq. (2-60) should be converted into Cartesian coordinates. Moreover, A,, A,, and A, may themselw be functions of r, 0, and z. In that case, they too should be converted into functions of x, y, and z in the final answer. The following conversion formulas are obvious from Fig. 2-13. From cylindrical to Cartesian coordinates:

y = r sin $J

z = 2 .

The inverse rilations (from Cartesian to cylindrical coordinates) are

, % ,

2 1

28 VECTOR ANALYSIS I 2 1 :

: t r 2 - , . r

Example 2-5 Express tbe yect& ' ,

4 = a,/3 cos 4) - ia$r + azs in Cartesian coordinates.

,.. Solution: Using

0

or

A = a,(3 cos2 q5 + Zr sin@) + a,(3 sin d, cos Q - 2r cos 4) + n,5: 1

But, from Eqs. (2-61) and (2-62),

and

sin 4 = Y . , Jq'

Therefore, , i

1 .I i

3x2 I , ,

1 ; which is the desired answer. i :

i

along the quarter-circle showp in Fig. 2- 14. I

" 1 ; J

, 1 : I i

I.

rig, f t14 Path for line integ~pl (Exgmple 2-6). - t

2-4 / ORTHOGONAL COORDINATE SYSTEMS 29 a

Solution: We shall solve this problem in two ways: first in Cartesian coordinates, k - then in cylindrical coordinates. f . : C

I a) In ~ a r t e & z n coordinates. From the given F and the expression for dC in Eq.

I (2-44), we have

I' F . dC= x y d x - 2x dy . L

it The equation or the quarter-circle ie x2 + y2 = 9(0 I x , y I 3). Therefore,

k ~ ~ - d t = ~ ~ ~ x ~ = ~ d x - 2 ~ ~ ~ ~ ~ ~ g I 1 0

= +g 3 - x 2 ) 3 / 2 1 - [ y ~ w + g s i n - ' q 1: 3

b) In cylindrical coorrlinates. Here we first transform F into cylindrical coordinates. n Inverting Eq. ( 2 -5), wc Il;lvc

I

i

> i I , With the given F, Eq. (2-63) gives L

i cos 4

0 I which leads to

T

-sin 4 cos (b (4

sin 4 cos 4 0

sin 4 cos 4 0

t F = a,(xy cos 4 - 2x sin 4 ) - a,(xy sin 4 + 2x cos 4) .

1 For ik-present problem the path of integration is along a quarter-circle of a

; radius 3. There is no change in r or z along the path (dr = 0 and dz = 0 ) ; hence 0- Eq. (2-51) sinlplifies to i dC = a93 d4 t

: and

. . 30 VECTOR ANALYSIS I 2

I

In this particular example, F is given in Cartesian coordinates and the path is circular. There is no compelling reasoli to solve tho problen~ in onc or the other coordinates We havc s l i ow~~ tlle co~lversin~l or v c c h ~ s i ~ n d the r occd~~ rc o(sollll\on i n both coordinates. r

over the surface of a closed cylinder about the z-axis specified by z = & 3 and r = 2, as shown in Fig. 2-15.

S o l i o ~ : In connection with Eq. (2-34) wc noted that the direction of L is normal to the surface. This statement is actuplly~imprecise because a normal to a surface can point in either of two directions. No ambiguity would arise in Eq. (2-35), since the choice of a,, simply determines the reference direction of currebt flow. In the present case, where F . ds is to be integrated over a closed surface (denoted by the circle on the integral sign), the direction of ds is always to be taken qs that of the oir~~vard normal. Our problem is to carry out the surface integral ,

O V l

1: t

the

Fig. 2-15 jA cylindrical Surface (Exapple 2 7 ) .

:ration. Along

ci the path is o r the othcr c 01 mlution

k I> 11or1n;il surface can

51. since the the present

he circle on .he out\c.rrrd

2-4 1 ORTHOGONAL COORDINATE SYSTEMS 31

over the entire specified surface. This integral gives the net outwardpw of the vector F through the enclosed surface.

, The cylinder in Fig. 2-15 has three surfaces: the top face, the bottom face, and '

the sidc wall. So,

..-.. We evaluate,the three integrals on the right side separately.

a) Top fa e. z = 3, a, = a,

F a, = k2z = 3k2 ds = r dr d+ (from Eq. 2-52);

lop F a, ds = C" So2 3k2r dr dq5 = 12nk2. face

b) Bottom fnce. 2 = - 3, all = -az

17 11,~ = - I ; ? : = 3 k 2 ds = r dr d4 ;

5;~,~l,,l,l F. i t , , ~ = 1 2 n / < ~ . I ~ L C

which is exactly thc samc as the integral over the top face.

c) Side wall. r = 2, a, = a,

ds = r d4 dz = 2 d 4 dz (from Eq. 2-52a);

F . a, ds = f:3 So2" k, d4 dz = 12nk,. side wall

Therefore,

$F ds = 12xk2 + 12nk2 + 12nk,

2-4.3 Spherical Coordinates

A point P(R,, dl, 4 , ) in spherical coordinates is specified as the intersection of the following three surfaces: aspherical surface centered at the origin with a radius R = R,; a right circular cone with its apex at the origin, its axis coinciding with the z-axis

VECTOR ANALYSIS / 2

Fig. 2-16 Spherical coordinates. -1 --

and having a half-angle d = d l ; and q half-plane containing the z-axis and making an angle 4 = 4, with the xz-plane. The pose vector aR at P is radialfrom the origin and is qzrite different /ram a, in cylindrical coordinates, the fatter being perpendicular to the z-axis. The base vector a, lies in the q5 F 4, plane and is t4pge&l to the spherical surface, whereas the base vector a, is the same as that in cylindrical coordinates. These are illustrated in Fig. 2-16. Far a right-handed system we have

Spherical coordinates are importanr 'for problems :involving point sources and regions with spherical boundaries. When an observer isvery far from the source region of a finite extent. the latter could be considered as the origin of a spherical coordinilte system; and, as a rcsult, suitiihlc simplifying ;~ppn)xim:ttions coi~J(l hc m;~clo. 'l.ili5 i* the rcason that spllcrical ioordinatesnrc i~rcd in solving an tcn~~a problems in thc far field.

Thc ' ent

ant

A vector in spherical coordinates iswritten as - '. , ~ F O I

mi

= a,AR -t a,A, + a A (2-65) or (

The expressions for the dot and crgss products of two vectors in spberi~al coordinates can be obtained from Eqs. (2-26) ~nd,(2-27).

In spherical coordinates, only R(u,) is a kngfh. ?he other two coordinates, B and 4 (u2 and u,), are angles. Referring to Fig. 2-17, w t q e a typical differential volume element is shown, we see that me t r i~ coekcients h, = Rand h3 = R sin B are required

k n g an

urr 'I. tc, .e >her; h a t e 3 .

2-64a)

2 -64b)

2 -64c)

:s and region dilute This is he fhr

2-65)

n nates

es, e ume lired

Fig. 2-17 A differential volume element in spherical coordinates.

'df = a, dR + a,R dO + a,R sin 0 (2-66)

The expressions for differential areas and d ikent ia l volume resulting from difler- entlal changes @, do, and dm in the three coordinate directions are

and

FO: convenience the base vectors, metric coefficient$ and expressions for the differ- entlal volume are tabulated in Table 2-1.

A ,VecfhgiWn in spherical coordinates can be transformed into one in Cartesian Or c~llndrlcal coordinates, and vice versa From Fig. 2-17, it is easily seen that

(2-69a) Y = R sin 0 sin 4 (2-69b)

(2-69c)

I . 1 t

34 VECTOR ANALYSISIF . ! ; . . , -

- , 1 \

I / I

I Table 2-1 Three Basic Orthogonal Coordinate Systems

<

Coordinate-system Relations a . --

P - Y-ctors.

- h 1

Metric Coefficients h,

Conversely, measurements in Cartesian coordinates can be transformed into those in spherical coordinates:

Cartesian Cylindrical Spherical Coordinates Coordinates . Coordinates

( x , Y, 4 (r3 47 4 (R, 8 , d - ax a, a~

a~ a, . Pz * a: a,

1 1 1 1 . r R

3 I 1 ' 1 i R sin 0 .

Example 2-8 The position of a point P in spherical coordinates is (8, 120°, 330"). Specify its location (a) in Cartesiali coordinates, and (b) in. cylindrical coordinatcs.

Solution: The spherical coordinates df the given poipt are R * 8; 0 = 120°, and 4 = 330". .- *

Differential dc Volume

a) I n Carresian coordinates. We use Eqs. (2-69a, b, c): !' : . '-4 ' , .. ! .

dx d y & ' r. dr d4' d; R 2 sin 0 dR dU dd

x = 8 s i n 1 2 0 ° c p ~ 3 3 0 " = 6 ,. . y = 8 sin 12 " &-'330° = - 2 8 ' '

r i P 1 .

z = 8 cos 120' F .-4. 1 ,

Hence, the location of the point is t ( 6 , 72J?, -4), and the position vector (the vector going from the origin to the point) is ,

1 . . 2 .

b) I1

' b! dl c;i

M

It coonl n; PC.

c n ~ r d , j

Esam ordin:

Solutl! This i a poir all po functic definir in gcn produ

Recgl; vector

0 - 6 9 ,

< \ -

3n

a,

a,

1 I-

R

R sin 0 i

I 0 d R dO d 4

f l in to *Iiosc

120', and i

0 v

1 > -

cector (the

2-4 / ORTHOGONAL Cf30RDINATE SYSTEMS 35 ' I ,

b) In cylindrical coordinates. The cylindrical coordinates of point P can be obtained by applying Eqs. (2-62a, b, c) to the results in part (a), but they can be calculated

a directly from the given spherical coordinates by the following relations, which can be verified by comparing Figs. 2- 11 and 2- 16:

r = R sin 8 (2-71a)

+ = ( b (2-/lb) z = Rcos0 . ( 2 4 1 ~ )

We have ~(4@,330", -4); anc. its position vector in cylindrical coordinates is

It is interesting to note here that the "position vector" of a point in cylindrical coordinates, unlike that in Cartesian coordinates, does not specify the position of thc point exactly. Can you write down thc position vector of the point P in spherical coordinates?

Example 2-9 Convert the vector A = a,A, + a,& + agAg into Cartesian coordinates.

a S o / f ~ ~ h ) ~ ~ : 111 111is J~I~OI)IC~II wc W~I I I to w r i t c A ~II l l i c rorm oSA ..- a, ,I , i- : I ~ / I , , -t ; I ~ . & I : . This is very diflerent from the preceding problem of converting the coordinate. of a point. First of all, we assume that the expression of the given vector A holds for nll points of interest and that all three given components A,, A, and ,A4 may bz functions of coordinate variables. Second, at a given point, A,, A,, and A, will have definite numerical values, but these values that determine the direction of A will, in general, be entirely different from.the coordinate values of the point. Taking dot product of A with a,, we have

Recalling that a, - a,, a , . a,, and a,. a, yield, respectively, the comDonent of unit vectors a,, a,, and a+ in the direction of a,, w e find, from Fig. i-16 and Eqs. (2-69a, b, c):

1. X - a, . a, = sin 8 cos 4 = pfy2f=:!

(2-72)

XZ a, a, = cos 0 cos + = (2-73)

J(x2 + y2)(x2 + y2 + z2)

Y :I,,, 11, = -sill (I x -. . - . --.-. (2 - 74) Js? +-j+

' . v

36 VECTOR ANALYSIS / 2 !

I ) I

Thus, ; i I

A, = 'AR sin 8 cos 4 + A, cos 8 cos ) - ,A, sip 4

Similarly,. i

AZ = A, cos 8 - A, sip 0 = A R ~ - - Jx2 + Y 2 + z2 J

A'm . . (2-77) xZ + y2 + z2

If A,, A,, and A, are themselves functions of R, 0, and 4, they t ~ o need to be converted into functions of x, y, and z by the use of Eqs. (2-70% b, c). ~ ~ S G o n s (2-75), (2-76), and (2-77) disclose the fact that when a vector has a simple form in one coordinate system, its conversion into apotber coordinate system usually results in a more complicated expression.

I )

Example 2-10 Assuming that q.cloud of electrons copfined in a region between two spheres of radii 2 and 5 ctqhag,q charge density of

' 4

find the total charge contained in the region. , . . ,

Solution: We have I

3 x 10-8 i = - ----1{4 -- c w 2 ,I),

. The given conditions of the pmbled obviously point to the use of spherical coordinates. Using lhc crprcssioo br du ia Eq. (2-6S), wc $crfoqn a triplc it~tcgration.

Q = So2$ Sn3'S0'O5 o! 0 .02 pR2 sin B ; ~ R dg d). I

Two things are of importance here. First, since p is giver in units of coulombs per cubic meter, the limits of integ~ation for R must be converted to meters. Second, the full range of integration for 0 ii from 0 to n radians, not from 0 to 271 radians. A little reflection will convince up that a half-circle (not a full-circle) rotatcd about thc z-axis

. I

r"" pos ' \-. CL, l.4

We at a are tior poi!

rep: OUS

dep line con cha tllc

1 coordi-

n. n

mbs per :ond, the ;. P, little hc :-axis

2-5 1 GRADIENT OF A SCALAR FIELD 37 '

through 2n radians (4 from 0 to 2n) generates a sphere. We have

0.05 1 Q = -3 x So2" S: So.o2 zcosz 4 sin 8 dR dB d+

2-5 GRADIENT OF A SCALAR FIELD

In electromagnetics wc have to dcal with quantities that dcpcnd on both time and position. Since three coordinate variables are involved in a three-dimensional space. we expect to encounter scalar and vector fields that are functions of four variables: ( r , i l l , 1 1 , , 1 1 , ) . In general, thc fields may change as any one of the four variables changes. We now address the method for describing the space rate of change of a scalar ficld at a given time. Partial derivatives with respect to the three space-coordinate variables i~ rc involvecl i ~ i ~ l . ~ I I : I S I I I I I C ~ I :IS ~ I I C r:~lc O I C I ~ ~ I I I E C I I I : I ~ ' I ~ C tlilBrc111 i l l tlillircnt ilirsc- (ions, a vector is needcd to delinc the space rate of change of a scalar field at a gibes point and at a given time. -

Let us consider a scalar function of space coordinates V ( L L , , u 2 , u,), which may represent, say, the temperature distribution in a building, the altitude of a mountain- ous terrain, or the electric potential in a region. The magnitude of V, in general, depends on the position of the point in space, but it may .be constant along certain lines or surfaces. Figure 2-18 shows two surfaces on which the magnitude of V is constant and has the values Vl and Vl + dV, respectively, where dV indicates a small change in V . We should note that constant-V surfaces need not coincide with any of the surfaces that define a particular coordinate system. Point PI is on surface V, ; P ,

38 VECTOR ANALYSIS 1 3 L

I /

is the corresponding point on surface Vl + dV along the normal vector dn; and P, is \ a point close to P, along another vector dP P dn. For the same change dV in V , the

, space rate of change, dV/d/, is obviously greatest along dn because dn is the shortest distance between the two surfa~cs .~ Since the magnitude of dV/dL depends on the direction of d€, dV/d& is a dipqional derivative. We define the vector that represents both the ?nagnitude and the dtrection of the maximum sppcc rate of increase of a scalar ' as the gradient of that scalar, We write

gradV !A a,, -. liV' dn I (2-78)

For brevity it is customary t~ employ tllc opcralor tlcl. rcprcscntcd by tllcsymbol V and write VV in place of grad 1'. Thus.

;

We have assumed that dV is positive (an increase in V ) ; if dV is negative (a decrease in V from P I to P,), V V will 6e negative in the a, direction.

The directional derivative along d€ is

dV dVdn dV - -=--- -- (te cln d/ dn

cos c?

- d 1' -r - a,. a, = ( V V ) . a,,

dn (2-80)

Equation (2-80) states that the space rate of increase of' y in the a, direction is equal to the projection (the compor)ent) of the gradient of V iq that direction. We can also writc Eq. ( 2 80)

(2-81)

where ti€ = a, df. Now, dVin Eq, ( is the total di rential of V as a rcsuit of a fFe change in position (from P , to P; 2-18); it Bn expressed in terms of the differentia1 changes in coordinates:

Eq. the

where d l l , d f , , and d t , are the components of the vector differential displacement a

d t in a chosen coordinate system. Ih terms of general orthogonal curvilinear coordi- ' i

In a more formal treatment, changes A V ' ~ O ~ A! would be used, and the ratio A V / N would become the ' der~vative dV/dC as hc fpproaches zero. We-avold this formality in f y o r of s~mpl~ctty.

. J

Loc

in ; defi

.... .;.,3.wi? -, ;;&+$;Fi;;;.i;va;,.;; .;;.,~&;.:&:y$%; . . . .". .. :,;; .. i , . > .

. . ' . , - ,, .

., . . - . > - .

' ., ,,.',<, f .

. . ;p $ P j; 2-5 1 GRADIENT OF A SCALAR FIELD 39

tnd P3 is in I/, the shortest

s on the rprcscnts fa sculur

nates (u, , u,, u,), dP is (from Eq. 2-31),

dP = a,, dd, + a,, de2 + a,, d t ,

= a U , ( h du1) + %,(h2 du2) + a U , h du3). (2-83) It is instructive to write.dV in E q . (2-82) as the dot product of two vectors, as follows:

av = + a,, - + a,, (a,, d l , + a,, d f2 + a,, di ,) ( , ae2

Comparing Eq. (2-84) with Eq. (2-81), we obtain

av av av V V = a u , - + a - + a -

a t , u2 a t 2 U3 ad3

I I

Equation (2-86) is a useful formula for computing the gradient of a scalar, when the scalar is given as a function of space coordinates.

In Cartesian coordinates, (u , , u,, u , ) = (x, y, z ) and h , = h , = h , = 1 , we have I 4

1 is equal C311 SO

In view of Eq. (2-88), it is convenient to consider V in Curreslun coordinutrs as a

W I ~ vector difkercntial opcrator.

Looking nt Eq, (2--Sh), onc is tc~np(ctl to dclinc V ;IS

I r coordi-

occoine the in general orthogonal coordinates, but one must refrain from doing so. True, this , , definition would yield a correct answer for the gradient of a scalar. However, the

. .. I I

40 VECTOR ANALYSIS I 2

same symbol V has been used cqnventionally to signify same differential operations of a vector (divq-gmcc and curl, which we will cansiber ]vier in this chapter), where

. an extension of V as an operator ip g~neral orthogonal coordinates would be incorrect.

Example 2-11 he electrostatic fiild intensity E i s derivable as the negative gradient of a scalar electric potential V; t h a t ' i ~ , ~ ~ = - VV. Determine E at the point (1, I, 0) if

'=Y . , a) V = Voe-" sip ;--, .. 4 I

b) V = VoR cos 9.

Solution: We use Eq. (2-86) to evaluate E = -VV in Cartesian coordinates for part (a) and in spherical coordinate6 for part (b).

'=Y = (ax sin 7 - ay cos y) vOe-%.

where I

In view of Eq. (2-17), the rpsult :above converts very'simply to E = - azVo in Cartesian coordinates. This is riot surprising sincena careful examination of the given V reveals that VoR cos 13 is, Fn fact, equal to V,r. 1n Cartesian coordinates.

216 DIVERGENCE OF A VECTOR F I C L ~ i i .

h . ,

of ' sci

by linc nit the

... . nor a fie1

wit the a nt

and (\I' I

m'o'' i

*J?t

,The the , surf. w h i ~ van that

tY PC ink: d~vic "PPr

n d i i

t

f "'x, , - .-. - differ can t

i 1 . In the preceding section we con6idqed the spatial derivatives of a scalar field, which led to the definition of the gradi~nr We now turn our attention to the spatial derivatives of a vector field. This will lead to the definitions of The divergence and the curl

. 1 I t I

I' , ' 1

.;,., ;L, ,i&$?.jr:p'.g u.. . * A ' . * @ t .c"zi- ,$.'. ;, .., . , ' , .. ,<,? . , , , ,,!.*:,. - , .- +,, y&.. *:-:. , ../ ' ,A;.!-:~ .:'::,' > i;! 3~; ' ;$5g#$g~&~&~~~g; ... . , . . . . . .- ', , ..{;;Y,$'',.:,$, *"2,7 , ..

' ' . ,%,.

., . , . . , .

2-6 / DIVERGENCE OF A VECTOR FIELD 41

p $5 -

perations f& r), where , @

ncorrect. . V

hrj

. gradient fi" 'f. >

{I, 1,O) if k

-a,VO in )I1 of the xdinates,

0 I

Id, which 1x1 dcriv- 1 the curl

of a vector. We discuss the meaning of divergence in this section and that of curl in Section 2-8. Both are very important in the study of electromagnetism.

In the study ofvector fields it is convenient to represent field variations graphically by directed field lines, which are called Jlux lines or streamlines. These are directed

, lines or curves that indicate at each point the direction of the vector field. The magnitude of the field at a point is depicted by the density of the lines in the vicinity of the point. In other words, the number of flux lines that pass through a unit surfxe normal to a vector is a measure of the magnitude of the vector. The flux of a vector field is analogov? to the flow of an incrinpressible fluid such as water. For a volume with an enclosed surface there 4 1 be an excess of outward or inward flow throu h

2 . - w y L t o w 2 .if!..,. the surficp only when tne volume contalns, respectively, a murce or a=, that IS, a net pos~tlvr, L,l v b -genre ; ~ x i i a ~ t e s the prrsence of a sourct of fluid inside the volume, ,aid ci net negative divergence indicates the presence of a sink. The net outward flow ' of the fluid per unit volullle is therefore a measure of the strength of the enclosed source. Qra& r

W e dejine the divergence of a vector field A at a point, -- abbreviated div A, as the net outwurdfiux o f A per unit volume as the volume ~ b o u t the point tends to zero:

$s A ds div A lim

AV-o AV '

The numerator in Eq. (2-go), representing thk net outward flux, is an integral over the entire surface S that bounds the volume. We have been exposed to this type of surface intcgral in Examplc 2 -7. Equation ( 2 -90) is thc gcncral dcfin~l~on of d ~ v A which is a scalar quuntity whose magnitude may vary from point to polnt as A itself varies. This definition holds for any coordinate system; the expression for div A, like that for A, will, of course, depend on the choice of the coordinate system.

At the beginning of this section we intimated that the divergence of a vector is a type of spatial derivative. The rc:tdcr niny pcrll:qx wondcr bout thc prcbcncc of ,\n integral in the expression given-by Eq. (2-90); but a two-dimensional surface integral divided by a three-dimensional volume will lead to spatial derivatives as the volu~ne approaches zero. We shall now derive the expression for div A in Cartesian coordinates.

Consider a differential volume of sides Ax, Ay, and Az centered about a point P(xo, yo, 2,) in the field of a vector A, as shown in Fig. 2-19. In Cartesian coordinates, A = a,A,-+a,& + a,A,. We wish to find div A at the point ( s o , yo, 2,). Since the dikrentii~l volumc has six fxcs, thc siirfucc intcgrnl in the nulncrator of Eq. (2-90) can be decomposcd into six parts.

42 VECTOR ANALYSIS / 2 . .

I

1 , Fig, 2 ~ 1 9 A differential volump in Cattesp coordi~:. ;-: ,

i

On h e front f a x , . 1.,,,,. A ' ds = Af,.;. As front = Afro., . a.(Ay Az)

face f a p face faoe

---. The quantity Ax([xo + (Ax/Z), YO, ZO]) can be expanded as a Taylor series about its value at (x,, yo, z,), as follows:

% + higher-order term4, (2-93)

where the higher-order terms (f4.q.T.) contain the factors AX}^)^, AX/^)^, etc. Similarly, on the back face, ' I . I

Ax d/l, = Ax(xU, yo, z ~ ) i- - 4~

YO, ZO) - - - Ax + H.O.T. (;-95) 2 'ax (sq, yo. a,)

'substituting Eq. (2-93) in Eq. (i-91) and Eq. (2-95); in ~ k . (2-94) and adding the contributions, we have I

( ~ 0 . YO. Z0)

Here a Ax has been factored out fromathe H.O.T. in Eqs. (2-93) and (2-95), but all terms of the H.O.T. in Eq. (2-96) still contain of A,+.

. ' . ', I > I

Hcr , face

91

whe Eqs.

The The evalt to ar

1

Exarr

( 2 92)

. ibout its

f l

( 2 -93)

~ / 2 ) ' , ctc.

( 2 -93)

. (2-95)

jdi;e?\h, (2 J )

j), but all

2-6 1 DIVERGENCE OF A VECTOR FIELD 43

Following.the same procedure for the right and left faces, where the coordinate changes are +hy/2 and -Ay/2, respectively, and b s = Ax Az, we find

Here the higher-order terms containthe factors Ay, (Ay)', etc. For the top and bottom faces, we have

, d

where the higkei-order'teims co'ntaiiibthe hctors A;; (Az)', etc. Now the results from. Eqs. (2-96), (2-Y7), and (2-98) are combined in Lq. (2-91) to obtam

+ higher-ordcr terms in Ax., A.Y, A;.

Since Aa = A x Ay A:, substitution of Eq. (2-99) in Eq. (2-90) yields the expression of div A in Cartesian coordinates

The higher-order terms vanish as the dillerential volume Ax Ay A: approaches zero. The value of div A, in general, depends on the position of the point at which it is evaluated. We have dropped the notation (x,, yo, z,) in Eq. (2-100) because it applies to any point at which A and its partial derivatives are defined.

With the vector diRerentia1 operator del, V, defined in Eq. (2-89) Ibr Currrsiao coordinates, we can write Eq. (2-100) alternatively as V A. However, the notation V . A has been customarily used to denote div A in all coordinate systems; that is,

We must keep in mind that V is just a symbol, not an operator, in coordinate systems other t h a q a r t e s i a n coordinates. In general orthogonal curvilinear coordinates (u,, u2 , u3), Eq. 12-90) will lead to

Example 2-12 Find the divergence of the position vector to an arbitrary point.

3 Solution: We will find the sqlution in Cartesian i s well as$n spherical coordinates. I

b

a) Cartesian coordinates. Thq expiession for thd position vector to an arbitrary .

point (x, y, z) is , QP=a,x.+a,y+a,z. (2-103)

Using Eq. (2-loo), we have . ' 1

f .

#

Its divergence in spherical conttlinatcs ( I < . 0, $)r';111 hc ohti~illcd fro111 tirl, ( 2 - 102) hy using Tablc 3- 1 as follows:

r

Substituting Eq. (2-104) in Eq. (2-105), we also 3btajn V (OF) = 3, as expected.

I d 1 8 ; 1-.,dA, 1 V - A = z z ( R 2 ~ R ) + - - - - - (A , yin 0) + P -. R sin 0 88 . R s m 8 84

Example 2-13 The magnet$ fluxl density B outpide a,very long current-carrying wire is circumferential and is ipversely proportional to t e distance to the axis of the wire. Find V . B. i ' I 1' . t

(2-105)

9 3 I . Solution: Let the long wire be coincident with the z-axif in a cylindrical coordinate system. The problem states that

, s , . 1 B=a, ; . k !

. . I .

In cylindrical coord in+s (r , 9, z), &. 12-102) r c d t k s lo . ,

1 i We have here a vector thqt is pgt a constant, but whose divergence is zero. This

property indicates that the mqgngtic flux lines close upqn themselves and that there are no magnetic sources or sinks. A divergenceless fielq is called a solenoidal field. More will be said about this t e btfield later m tbe boqk.

t YP

)ordinates. i' d 2..

1 arbitrary $. . ::.

(2 - 103) \ \

t -I

that there 7idal field.

2-7 1 DIVERGENCE THEOREM 45 ,

,

Fig. 2-20 Subdiv~di volurne for proof of divergence theclrzrr,, :

2-7 DIVERGENCE THEOREM

In the preceding section we defined the divergence of a vector field as the net outward . flux per unit volume. We may cxpect inluitivcly that the uolwnr inteprul of ,/he ihrr - gence of a oecror field equals the total outward f lux of the vector through the surjncr that botolds the volun~c; that is,

' I liis i d c ~ ~ l ~ l y . wl~ich will he p.ovsd ill ihc li)llowillg iii~li~g~.iipll, is ca11cd the ~ ~ I I ~ , , O C ~ I C L ,

theorem.' It applies lo any volume V that is bounded by surface S. The direction of ds is always that of the outward normal, perpendicular to the surface ds and directed away from the volume.

For a very small differential volume element Auj bounded by a surface s,, the definition of V A in Eq: (2-90) gives directly

- -, In case of an arbitrary volume V , we can subdivide it into many, say N , small differential volumes, of which Auj is typical. This is depicted in Fig. 2-20. Let us now combine the contributions ofall these differential volumes to both sides of Eq. (2-108). We have

1 [ (V A), A , = 1 [ A d . (2-109) A u J - 0

j = 1 A v J - 0 ----.- j = 1

The left side of Eq. (2-109) is, by definition, the volume integral of P . A:

It is also known as Gauss's theorem.

' . ? ' 1

46 VECTOR ANALYSIS I 2 !, i ! \ !

5 4 .

The surface integrals on the right side of Eq. (2-109) are summed over all the faces of all the differential volume elements. T-he contribdtibns from the internal surfaces of adjacent elements will, however, cancel each other; bequ+ at a common internal : surface. the outward normals of the adjacent elemehts point in opposite directions. Hence, the net contribution of the right side of Eq. (2-109) is due ~ i d y to that of the external surface S bounding the volume V; that k, ,

i lie suustnuuon of Eqs. (2-110) and (2-111) in Eq. (2-109) yields the divergence theorem in Eq. (2-107).

The validity of the limiting processes leading to the proof of the divergence theorem requires that the vector field A, as well as its first derivatives. exist and he continuous both in I/ and on S. Tllc:divcrpmx thuo~.cnl is ;ttl inlport:~~lt idc11t1(y 1 1 1

vector analysis. It converts a volimc intcgral of thc diwrgcncc of a vector to a cluscd surface integral of the vector. apd rice versa. We use it b e q t e ~ t l y in establishmg . other theorems and relations in electromagnetics. We note +at, although a s~ngle integral sign is used on both sides of Eq. (2-107) fot simpticity, the volume and surface integrals represent, respectivciy, trlple and do!~bla, integrations.

Example 2-14 Given A = a,*' + a p y + a , p , verify the*divgrgence theorem over a cube one unit on each side. The cube is situated in the first octant of the Cartesian coordinate system with one corqer at'the origin. i

1 I

Solution: Refer to Fig. 2-21. Wg fir<t :valuate the &face jntegral over the six faces , , 2 , -

- r

face r

2. Back face: x = 0, cis = -a, 4 y @; , t

. . i ' A - d s = O . ( back face

! ' . L . .

1; ' ,

U . \ I , . I , .. . ,

., - !' ! \.: : , .

' ; 1. : . , y

Fig. 2-21 A j l i ~ i t cube I . . r ). ,

I' . (Example 2-14):. . 5

, . 5 5

Fig. 2-21 A'uhit cube I .

(Example 2-14): , ). 5 5

Henc

as be

Exar fort Cdl l l<

n

, , / '

1;7 .!. 2-7 1 DIVERGENCE THEOREM 47 -.

[;i I the faces i.5 3. Left face: y = 0, ds = -a, dx dz ; 41 surfaces r ' 111 internal $ Lf, A - d s = 0. iirections. c face

to that of ! B$ $7'6 ' 4. Right face: y = 1, ds = a, dx dz ; it;, A ds = SO'So' x d x d z = $. ! ngh t

(2-111) ' Y face p. 5. Top face: z = I, ds = dx dy a,; livergence F

k J. -h A ds = JolJo' y d x dy = 4.

& face

v . J

iivergence s t and be 6. Bottom face: z = 0, ds = -aZ dx d y ;

dentity in o a closed Lo.., A . ds = O.

~ > I I \ I I I I I ~ face

! a mglr Adding the above six values, we have

IL1"Pnd ( F . \ - ' / s = I + o + o + ~ + ~ + o = ? .

Now the divergence of A is ~ m n over

Hence, six faces.

J v V s .A dv = So1 So1 So1 (3x + y) d x d y dz = 2, .- as before. $

Example 2-15 Given F = a,iR, determine whether the divergence theorem holds for the shell region enclosed by spherical surfacs at R = R , and R = R,(R, > R,) centered at the origin, as shown in Fig. 2-22.

Fig. 2-22 A sph&ical shell region (Example 2-15).

1 i t i '- 48 VECTOR ANALYSIS / 2 I

$ -

1 I I '

i I * 7 1

Solution: Here the specified region has two surfaces, pt R = R , and R = R 2 .

At outer surface: R = R,, $s = a R ~ : sin 0 dB d&;

I.., F ds = Joan 1; (kR,)R: sin B dm = 47ckR:. surface

Actually, since the integrand is independent of 0 or.# in both cases, the integral of a constant over a spherical surface is.simply the constant multiplied by the area of the surface (47rR: for the auter sl~rfa$e and 4 7 r ~ : for the inner surface), and no intcgra- tion is necessary. Adding thc two results, we have.

---. To find the volume integral we first determinev . F for an F that has only an

F R component: , :

Since V . F is a constant, its voluqe integral equa1s:the product of the constant and the volume. The volume of the shall region betweed the two spherical surfaces with radii R, and R 2 is 47r(~3, 1 ~ : ) / 3 . Tkrefore, !,

as before. i r

This example shows that the divergence t h e o r holds even when the volume has holes-that is, even when the vplume is enclosed!by a rriultiply connected surface.

: s A .

1 : I

2-8 CURL OF A VECTOR FIELD , I . I In Section 2-6 we stated that a net outward flux &f a vector A through a surface bounding a volume indicates the p&ence of a souqce. This source may be called a

. pow source and div A is a measure of ihe strength of ;he flow source. There is another kind of source, called vortex source, which causes a cipdat ion of a vector field around it. The net circulation (or simply circulation) of a veqor. field around a closed path is

I defined as the scalar line integrql of the vector over &e,p<~h. We have

i . I Circulation of A arbund contour I$ $c A . dt?

i . 1,

Equation (2-1 12) is' a mathematical :qefinition. The meaning of circulation depends on what kind of field (be vector A represents. If A is a force acting on an object, its circulation will be thg wor done by the force jn moving the object once

" F ; . ! , { . +

aro will boc of L'\

9 kalr I

SOU

a rr

In v x (1.7 t

the arc; the dirt cL'1:

nvecr ~ t h ~ lmi:

whe AS"

e volume d surface.

a surface 3 called a s another

rci~l;~tion ng on an ject once

2-8 1 CURL OF A VECTOR FIELD 49 .

\ \

Fig. 2-23 Relation between a, and dP in defining curl.

awund the contour; if A represen .s an I-::::t:ic field intensity, then the circulation wil! be an electromotive force ;,uund the closcd path, as we shall sce later in the. book. The familiar phenomenon of walcr whirling down a sink drain is an suarnplc of a vortex sink causing a circulation of fluid velocity. A circulation of A may exist even when div A = 0 (when there is no flow source).

Sincc circulation as dclincd in Eq. (2-1 12) is a line integral of a dot product, its value obviously depends on the orientation of the contour C relative to the vector 4 . In order to define point function, which is ;L ii~e;lsosc o i llic strength of2 vbrtea source, we must make C very snlall and orient it in such a way that the circulation is a maximum. We definet

In words, Eq. (2-113) states that ;he curl of u vector field A, denoted b~ Curl A or V x A, is a vector whose nlugnitide is the maxin~am net circ~dation o f l per unit urea as the area rends ro zero and whose direction is the normal direcrion of the area w ! m rhe area is oriented to make the net circulation rna.uirnlu?l. Because the normal to an area can point in two opposite directions, we adhere to the right-hand rule that when the fingers of the right hand follow the direction of dP, the thumb points to the a,, direction. This is illustrated in Fig. 2-23. Curl A is a vector polnt funct~on and 1s conventionally written as V x A (del cross A) although V 1s not to be consdercd :1

vector operator except in Cartesian coordinates. The component of V x A in any other direction a, is a, . (V x A), which can be determined from the circulation per unit area n q a l to a,, as the area approaches zero. -

(V x A),, = a,, . (V x A) = lin1

whcrc [IN tlircction o r Ihc liw in(cgr:i~ion around t l ~ c contour C, bounding m a As, and the direction a, follow the right-hand rule.

' In books published in Europe the curl of A is often called the rotation of A and written as rot A

i.

50 VECTOR ANALYSIS / 2 ~, i 1;

1 * 1 $ 4 \ 1

1 1

I

I

I

1

- . .- -. --J, Fig. 2-24 Detemlning (V x A),.

We now use Eq. (2-114) to find the three components of V x A in Cartesian coordinatca. Ilcfcr to i:ig. 2 -24 wllerc :I dillcsc~il i : ~ l I .CCI:III~III :I~ :I~C:I p:~l.i~ll~l 1 0 tllr y:-pIanc a11d h;~ving siclcs A!, [~nd A: is ~ ~ : I \ V I I ; IIIOIII :I ( ) / J ~ C ; I I 1)c)i11t I ~ ( . Y ~ ~ , jaf,, :,,),

We have 3. = aT and As. = Ay LIZ and the contour C. m ~ s i s t s of-thelour sides l ,2,3, and 4. Thus,

Ay Az-0 Ay Az (2-115)

1, 21'3. 4

In Cartesian coordinates A = a,A.=,+ a,A, + a,A, The contributions of the four sides to the line integral are

I "' 3 . x,, yo + T , 3, can be expanded as a ahl lor series:

I I ' .

L -

:artesian Icl to the , I j.0, 4. cs I , ? , a,

r) (2-" 5 )

l l ~ r I L ) L I ~ ~

(2-116)

(2-117)

-0

(2-118)

(2-119)

2-8 1 CURL OF A VECTOR FIELD 51 ,

Note that dP is the same for sides 1 and 3, but that the integration on side I is gcing upward (a Az change in z), while that on side 3 is going downward (a - Az change . in z). Combining Eqs. (2-117) and (2-119), we have

(2 - 120) l k 3

The H.O.T. in Eq. (2-120) still contain powers of Ay. Similarly, it may be shown .hat

( + H.O.T.)~ Ay Az. LMc, A dt = ---?

az (2--121) 2 & 4 Ih. Yo, 2%)

I .

Subsriru~'ng Eqs. (2-120) and (2-121) in Eq. (2-115) and notlng that the hlgher- order terms tend to zero as Ay + 0, we obtain the r-component of V x A :

dA.. dA , ( V X A ) =-..-z---!. (2-123) " dy dz

A close examinat~on of Eq. (2-122) wili reveal a cycl~c order in 1. y. and : and enable us to write down the y- and ;-components of V x A. The entlre expression 1'0s tllc curl ol' .\ in C,~r lcs~a~l c .oor~l in,~(c~ 1s

dA* dAy V x A = a , -2-- + ayt$ - $I.+ a=(? - $1. (2-123)

7 I-.--

Comparcd to lllc cxprcssion for V . A ~n Eq. (2-IOO), that for V x A ~n Eq. (2-123) a morc complicalcd, as it is expected lo be, bcciiuse 11 1s a vector w ~ t h lhrce con~ponents, whereas V . A is a scalar. Fortunatciy Eq. (2-123) can bc remembered rather easily by arranging it in a determinantal form in the manner of the cross product exhibited in Eq. (2-43).

The derivation of V x A in other coordinate systems follows the same procedure. However it is more involved because in curvilinear coordinates not only A but also

' -A- dP changes m magnitude,as the integration of A - dP is carried out on opposite sides of a curvilinear rectangle. The expression for V x A in general orthogonal curvilinear coordin;~tes (II, , n2. l r l ) is given l x h v .

52 VECTOR ANALYSIS / 2 1 :

. \.

It is apparent from Eq. (2-125) td4t an operalor k r m cannot be found hcre for the symbol V in order to consider 8.x A a cross pl;oduc(, The expressions of V x A in cylindrical and spherical k8or;Bdates can be eqily obtained from Eq. (2-125) by using the appropriate u,, u,, q d y, and their metric coefRcien!s h,, h, , and h,.

*'.. '

Example 2-16 Show fhat V x A 1 0 if I

a) A = a,,,(k/r) in cylindricalcpordjnntc~, whcrc k:is n canstanlt. or

b) A = a , f ( R ) in sphepcal cpordinates, where f ( R ) is a n y fuqcion of the radial distance R.

Solut iorz , a) In cylindrical coordinates th; following apply: (u , , a,. r,) = (r . 4,:):. b , = 1.

11, = 1.. and 1 7 , = 1. Wc havc. from Eq. ( 2 .135).

i I 1 / ir

which yields, for the given A, , .

I t b) In spherical coordinates the following apply; ru,, u, , u,) = (R. 0.4): h , = I h, = R, and h, = R s iq ) . Hence. . .

1 > . V X A = - - R 2 sm 9 dR

i A ,

and, for the given A, I

I

' c for thc L > f V x A - 125) by 3 .

2-9 / STOKES'S THEOREF. 53

A curl-free vector Iield is called an irrotutiotxd or a conservative field. We will see in the next chapter that an electrostatic field is irrotational (or conservative). The expressions for V x A given in Eqs. (2-126) and (2-127) for cylindrical and spherical coordinates, respectively, will be useful for later reference.

2-9 STOKES'S THEOREM

For a very small differential area As, bounded by a contour C,, the definition of V x A in Eq. (2-1 13) leads to

f

(V x A), . (As,j = ( A . ( I t . .lc'

( 2 - 128)

In 0ht:iiung Eq. (2-i28), we have taken the dot product of both sides of Eq. (2-! 13) with a,, Asj or Asj. For an arbitrary surface S, we can subdivide i t into many, say :\I,

small differential areas. Figure 2-25 shows such a scheme with Asj as a typical differential element. The left side of Eq. (2-128) is the flux of the vector V x A through the area Asj. Adding the contributions of all the differential areas to the flux, we have

N - ,. lim ? ( V X : ~ ) , ~ . ( A ~ , ) = ( V X A ) - ~ S . \s, . o 3

.i I J . Y

Now we sum up the line integrals around the contours of all the differential elements rcprcscntcd by thc rig111 sitlc of Eq. (2- 128). Sincc the common part of thc contours ol' two i~tl.i:~ccnl clclnc~~ls is I~.;\vcrsctl in oplxwiilc tlircctio~is by two contours. thc ncl c o t ~ l ~ i l ) t ~ t i ( ~ ~ ~ or:\ll t l \c C I ) I I I I I > I I I I 1)iII'LS i l l I I K i~thxior to ~ I I C lokt1 line in~cgriii is x ro , and only tht: contribu~ion I'rom he external contour C bounding the entire area S remains after the summation.

Combining Eqs. (2-129) and (2-130), we obtain the Stokes's theorem: .

/ Js (V x A ) . ds = $= A . dP, 1

Fig. 2-25 Subdivided area proof of Stokes's theorem.

for

54 VECTOR ANALYSIS ! 2 . ' I

4 . ?

which statcs that the surfice i r y r d u i the curl oj:o aw#ur jield over wr upen surjuce is equal to the clpsed line integl;al q{!the vector don$ the contour bounding the surface.

As with the divergence t h e o r i d the validity of the limiting processes leading to the Stokes's theorem requirei thqi the vector field A, as well as its first derivatives, exist and be continuous both on S and along C. Stokes's theorem converts a surface integral of the curl of a vector t~ hiline integrai af tlre vector, and vice versa. Like the divergence theorem, Stoke+ thdorern is an imdortadt identity in vector analysis, x d we will use it frequently in sther theorems and relatrons in i.! *' r, aagnetics. ,

If the surface integral of V x A is carried over a tlosed surface, there will be no surface-bounding external contour, and Eq. (2-131) tells us that

g i p x A ) - & = o (2-132)

for any closcd surhce S. The pcmctry in Fig. 2-'25 is c h ; ~ i l dclibei:llcly to ~111-

phasize the fact that 3 nontrivial application of Stvk~s 's Lhcoccn~ s l ~ n y s implies ail

open surface with a rim. The simplest open surface would be3wo-dimensional plane or disk with its circumference as the contour. We remind ourselves here that the directions of dP and ds (a,) follow the right-hand rule.

Example 2-17 Given F = a,ry - a,.2x, verify SSLkes's theorem over a quarter- circular disk with a radius 3 < i n the first quadrant, as was shown in Fig. 2-14 (Example 2-6). , : : I

I ! '

S i : Let us first find the (urfaje integral of V:x F. Prom Eq. (2-130).

Therefore,

(2 - sur C 1

I w / w c surfice.

d i n g to ivatives, L surface xi. Like

Jons in I C

ill bt no 4 ',.

2-10 1 TWO NULL IDENTITIES 55 !

It is itnporlun{ lo usc thc: propcr h i t s for thc two variables of inlcgration. Wc citn interchange the order of integration as ,

and get the same result. But it would be quite wrong if the 0 to 3 range were used as the range of integration for both x and y. (Do you know why?)

For the line integral around AB0'4 we hdve already evaluated the part around the arc from A to B in Exarilple 2-6.

From R to 0 : x = 0, and F : de,- F . (11, ~ ' y ) = --'2% dy = 0.

From 0 i o A': y = 0, and,F. dP = F . (a, dx) = r y dx = 0. Hence,

from Example 2-6, and Stokes's theorem is verified.

Of course, Stokes's theorem has been established in Eq. (2-131) as a genzral identity; there is no need to use a particular example to prove it. We worked out the example above for practice on surface and line integrals. (We note here that both the

, vcclor ficltl i ~ n t l ils lirst spalii~l ilcrivalivcs nyc linitc ancl co'ntinuous on the surface ah wcll xi on LIIC contour 01'in~crest.)

2-10 TWO NULL IDENTITIES

Two identities involving repeated del operations are of considerable importence in the study of electromagnetism, especially when we introduce potential functions. We shall discuss them separately below.

I

2-10.1 Identity I

In words,@ curl of the gradient of m y scalar Jield is identically zero. (The existence of V and itsfitst derivatives evcrywhcre is implied here.)

Equation (2- 133) c;ln he p r a \ d sc:liiily i n Car~csi;~n coordin:~tes by usmg 51. I;! 39) I'oI' \.' :lntl 1x1 l ; ~ I I I I I I ~ Ll~c i ~ r c l u k d C)I)CI ' :L~~OIIS. 11) SCIICIYII , il. wc UJ\C 111~'

s'urface integral of V x (VV) over any surface, the result is equal to the line integral of V V around the closed path bounding th i surface, as asserted by Stokes's theorem:

Ss.[v x (VV)] . '1s = (2-1 34)

I I

56 VECTOR ANALYSIS / 2 1

I '

a , 1 . ,'I + ? . 1

However, from Eq. (2-81), , I I

The combination of Eqs. (2-134) and-(2-135) states'that the surface integral of V x ( V V ) over any surface is zero. The integrand itself must therefore vanish, which leads -+ to the identity in Eq. (2-133). Since 4 coordinate system is qot specified in the deriva- t iox the identity is a gencral Qne a i d is invariant jwith the choiccs of coordinatc

I i t . q; stems. F

4 converse statement r :' Identity I can be mad$ as fqllows., I f a .:tor jeld is ; , , * curl-free, then it can be expressed as the grudient of a scalar J'ieid. Let a vector field

be E. Then, if: V x E = 0, we can define a scalar field V such that

I

The negative sign here is unimportant'as far as Identity I is concerned. (It is included in Eq. (2-136) because this relation conforms with a' basic relation between electric field intensity E and electric scalar potential V in electrostatics, which we will take up in the next chapter. At this stage it is immaterial what E and V re'$aent.) We know from Section 2-8 that a curl-free vector field is a conservative field: hence an irrotational (a conservative) vector jiell( ccu; ulways he expressed as the gradient of a scalur field.

I

i '

2-10.2 Identity ll i n fio !I L

I

In words, the divergence o j the curl u f g n y vector ~ i e l i i s identically zero. u r or

Equation (2-137). too, can be ~ r b v e d easily in Cartesian coordinates by using a r Eq. (2-89) for P and performing $he indicated operatipns. We can prove it in general without regard to a coordinate syst& by taking thq'volurne integral of V . (V x A) PC

on the left slde. Applying the divcrgqncc thcorcm, wu.havc 2-11 HE

CII Let us choose, for example, the arbitrary volume v e n c ~ o s e d b ~ a surface S in Fig. 2-26.

. The closed surface S can be spli) into two open surfaces. 9, and S,, connected by a n bc

common boundary which has bean dkywn twice as CIthrrd G,. We then apply Stokes's 1 theorem to surface S, bounded by C,; and surfacc SL boundcd by C,, and write the right side of Eq. (2-138) as . I 1, I

; \ 1

fS (V x A) . ds = Js, (v.! A) an1 b + Js (V r A) . a,,, cis 2

. ,- , , . .

! . . . , .: . 2 .', . -

2-11 / HELMHOLTZ'S THEOREM 57

(2-135)

11ofV x ~ c h leads : deriva- ordinate

r jkki :tar field

(2-136)

,nciuded I clxtric I take up i c l a m

,, \ :. lJ0;

(2 137)

2y using I general tv x A)

(2-138)

lg. 2-76. red by a Stokes's v r i t n

(2-139)

1 1

Fig. 2-26 An arbitrctty volume I/ enclosed by surrace S.

The nor-.als a,, and a,, to surfaces S, and S, are q:tward normals, and their relatioiia with the path directions of C, and C2 follow the &&-hand rule. Since the contours.

' C, and C, are, in fact, one and the same common boundary between S , and S,, the two line integrals on the right side of Eq. (2-139) traverse the same path in opposite directions. Their sun1 is therefore zero, and the volun~e integral of V - (V x A) on the left side of Eq. (2-138) vanishes. Because this is true for any arbitrary volume, the integrand itself must be zero, as indicated by the identity in Eq. (2- 137).

A converse statement of Identity I1 is as follows: I f a vecror field is r l i l ;e r ;~r i lc~ l i .ss . rlr('rr ir cqtrlr Iw c s p ~ ~ s s c ~ t l 11s (Irl, cir1.1 o / ' t r i ro t l~c i . UL~L ' IOI* j i ~ l t l . Lct ;I vcctor held be B. This converse statement asserts that if V - B = 0, we can define a vector field A such that

Ii -: V x A . ( 2 - 140) In Seclion 2-6 we mcntionccl th:~t ;i divcrgcncclcss licld is also called a solenoidal ficld. S ~ l c n ~ i ~ l a l ficltls ;lrc nol :~ssoci;~~c(l will1 I low s o ~ ~ ~ . c c s 01. s i ~ ~ h s . Tllc net outw:~rcl llux of a solcnoidul licld L I I I U L I ~ I I :IIIY closed surl'acc is zcro, 2nd the l l ~ ~ x lincs ciosc upon themselves. We are reminded of the circling magnetic flux lines of a solenoid or an inductor. As we will see in Chapter 6, magnetic flux density B is solenoidal and can be expressed as the curl of another vector field called magnetic vector potential A.

2-11 HELMHOLTZ'S THEOREM

In previous sections we mentioned that a divergenceless field is solenoidal, and a curl-free field is irrotational. We may classify vector fields in accordance with their being solenoidal and/or irrotational. A vector field F is

1. Solenoidal and irrotational if 1. V . F = O

and V x F = O .

Iistr~rrpl~~: I\ sutic clccwic l i d t l ill :I charge-li.ccl rcgian.

*2. Solenoidal but not irrotational if

V . F = O 'and V x F f O .

Example: A steady magnetic field in a current-carrying conductor.

58 VECTOR ANALYSIS I 2 I

- .

3. Irrotational but no,t solenoidal if

V'x F = O and V - F f O .

Example: A static electric field in a charged region.

4. Neither solenoidal nor irrotationnl ii ,

V - F # O ' rr.2 V X V # ~ .

E x a m p l e An electric field in ; X' - r t ; ~ .L rnw!i~ In, it!! .i finre - ~ ~ ~ ; r y m g inagletlc field.

The most general vcctor licld thcn has holh-a nonzero divcrgcncc :)lid :I wnzcro curl. and can bc considcrcd as rhc sum ol'u solcl~oitI:~l licld : ~ n r l 311 irro(;~iiol~;tl licld.

Hclrl~holr:'.~ Tlrc~o,wn: rl vector ~ i c / t l ' . ( ~ w ~ t o r poirlt Jrrrlctior~) is ~I~t~~rlr irrc~d to \\Ythin (211 udditiw co~~slail t I / ' both its diwyerzce am/ its curl ure specijed everywlzer-k. In an unbounded region we assume that both the divergence and the curl of the vector field vanish at infinity. If the vector field is confined within a region bounded by a surface, then it is determined if its divergence and curl throughour-the region, as well as the normal componeht of the vector over the bbundirlg surface, are given. Here we assume that the vector'functfon is single-valued and that its derivatives are . , . . finite and continuous.

Helmholtz's thcorem can be proved as a rnathcmaticali~hcorcm iq a gencral wayt For our purposes, we remind pupelves (see Section 2-8) that the divergence of a vector is a measure of the strenith of,the flow source and.that the curl of a vector is a measure of the strength of t heg~g tex Source. When the strenphs of both the flow source and the vortex source a'% 'specified, we expect that the vector field will be determined. Thus, we can decompose a'gineral vector &ld F-intoran irrotational

6 .

part F, and a solenoidal part F,: 3:.: , .: . A ' .

with

and

F = Fi + F,,

where g and G are assumed to be known! We have ,

I .* I

V . F = V y F , = g 1 . I (2- 144) and

I . I \ C x F = ' i x F , = G . ,, (2- 145)

Hrlnlhdtz's theorem asserts that when 4 and G i r e sFc&d. the vector fmction F

the :

Exa

. , See. ior instance. G. Xnlen. Afarhemoricaf Jferlt@s for Phydcisrs, .+idemic Press (1966). Section 1.15.

. .

2-11 1 HELMOHLTZ'S THEOREM 59

,: h is determined. Since V . and V x are differential operators, F must be obtained by $3 ?j integrating g and G in some manner, which will lead to constants of integration. The is determination of these additive constants requires the knowledge of some boundary i? , conditions. The proccdurc for obtaining F from given g and G is not obvious at this t time; it will be developed in stages in later chapters. !t The fact that Fi is irrotational enables us to define a scalar (potential) function

V, in view of identity (2- M ) , such that i . P 1:

F, = - V V . p 1 1 6 )

Sirnihdy, identity (?-137) 2nd Eq. (2- l43a) allow the definition of a vecLc; (potential) function 4 such the?

Helmholtz's theorem states that a general vector function F can be written as the sum of the gradient of a scalar function and the curl of a vector function. Thus.

F = - V V + V x A . (2- 14s)

In following chapters we will rely on Helmholtz's theorem as a basic eiemznt in the axioinatic development of electromagnetisnl.

Example 2-18 Given a vector function

I: - $ I , ( 'I\, - 0 , T I I ; t , , ( < , 2 \* .-. 2:) -. :I ( 1 , , I , 1. z ] ,

3) Oe~ennine the constanls c,, c2 , and c3 il. 1: is irrolaiional.

M Determine the scalar potential function V whose negative gradient equals F.

Solution a) For F to be irrotational, V x F = 0; that is,

Each-wqlponent of V x F must vanish. Hence, c , = 0, c2 = 3, and e, = 2.

b) Since F is irrotationnl. i t cnu be e\presscd as the negitlve gradlent of a sc.ll;lr I'u~iclion I:; 111:~ is,

i. i.v av av F = - V V = -ax -'- ay -;-- - a= -

ax oy az = ax3y + ay(3x - 22) - a1(2y + z ) .

60 VECTOR ANALYS 1:

1 ' Three equations are obtained: r . > L .

av' ' t T

-= -3y a . ~ (2- 149)

Integrating Eq. (2-149) partially with respect to x, we have

I' = - 3sy + ,f',(,Y. z) . 2

(221 5 2 )

and

Examination of Eqs. (2-152), (2--1.53); and (2-154 enables us to write the scalar potential function as

- Any constant added to Eq. (2- 155) would still mkke lran answer. The constant is to be determined by a boundary condition or the condition at infinity.

REVIEW QUESTIONS I

R.2-1 Threc vectors A. B. ;md C. drqan ib .I i~eod-to-tail f.l{hion. f p n ihrce sidcs of :I tri;~n$e. What 1s A f B + C'? A + B - C ? , , . . R.2-2 Under what conditions can the dot product of two vectors be negative?

I R.2-3 Write down the results of A - B aridlA x B if (a ) A I( I$, and fb) A 1 B.

R.2-J Which of the following produkts of vectors do not mqke sense? Explain.

a ) ( A . B) x C b) A@ C) , *

c ) A x B x C d l A/B c) Ailan f ) (A x p 1 . C

. . - . >::;:.. . . ,: '.:>J$.1.:.' . ,, .' . , . + . ., . . , . ' ,. . . .< > i :.>$J. !.

.> ., * ; -., , ,4,; .; ,,

. . !$ REVIEW QUESTIONS 61

R.2-5 Is ( A . B)C equal to A(B . C)?

R.2-6 Does t i . B = A C imply B = C? Explain.

R.2-7 Does A x B = A x C imply B = C? Explain.

R.2-8 Given two vectors A and B, how d o you find (a) the component of A in the direction of B and (b) the component of B in the direction of A?

R.2-9 What makes a coordinate system (a) orthogonal? (b) curvilinear? and (c) right-handed?

R.2-IC Given a idctor F in orthogonal curvilinear coordinates (u , , u,, u,), explain how to determin, (a) F an2 (b) a?. #

11~2- 1 1 WhzL arc i ixlL. ic c d ~ i c i ~ , , h :

' R.2-12 Given two points P l ( l . 2. 3) and P 2 ( - 1, 0, 2) in Cartesian coordinates, write the espres- sions or the vectors 1 7 : ;111d /TI. R.2-13 What are the expressions for A . B and A x B in Cartesian coordinates?

R.2-14 What are the values of the following dot products of base vectors?

:I) i lp ;I., 1)) :Ir . :Iy

c) :I,< :hr d) a , . ;I, e) a, a, f ) ar az

R.2-15 What is thc physicid dcfinition orthe gradicnt of a s c ~ l a r field?

11.2-10 Exprchs tllc space ralc ofchiunyc of a scalar in u yivcn direction in tcrms of its gradicnt.

11.2-17 What does the del operator V stand for in Cartesian coordinates?

R.2-18 What is Ihc physical dcfinition of thc divcrgcncc of a vcctor ficld?

R.2-19 A vector field with only radial flux lines cannot be solenoidal. True or false'! Explain.

R.2-20 A vector field with only curved flux lines can have a nonzero divergence. True or false'? Explain.

R.2-21 State the divergence theorem in words.

R.2-22 What is the physical definition of the curl of a vector field?

R.2-23 A vcctor ficld with only curvcd flux lincs cannot bc irrotational. Truc o r Salsc'? Esplain.

R.2-24 A vector field with only straight flux lines can be solenoidal. True or false? Explain.

R.2-25 StakS&o_kes's theorem in words.

R.2-26 What is lhc dilkrcncc bclwccn a11 ilmtation:il field nud a solcnoid;il field?

1$2-~27 Slillc I l r )h , ( l ; ' s \lrco~>cln in wcw~la.

R2-28 Explain how a general vector functiomcan be expressed in terms of a scalar potential function and a vector potential function.

1

62 VECTOR ANALYSIS I 2 i

3 i L

PROBLEMS

P.2- 1 Given three vectors A, B, and C as follows, C

A = a, + a,2 - a,3 B =,-a,4 + a, -

find

e ) the component of A in the direction of C f ) A-; C g) A . ( B x C)and(A x B ) - C 11) (A x B) y C and A x (B x C)

P.2-2 The three corners of a triangle are at P,(O. 1. -2), P,(4, 1, -3), and P,(6, 2, 5). a

3) Dctcrn~iiic \vl~c[l~cr A PI lJ21'., is i t ri$it [ V ~ : I I I ~ I ~ - . b) Find the area of the triangle.

P.2-3 Show that the two diagonals of a rhombus i r e perpendicqlar to e ~ h g _ t h e r . (A rhombus is an equilateral parallelogram.)

P.2-4 Show that, if A B = A C nr~d A x B = A x C, where A is pot a null vector, then B = C.

P.2-5 Lnlt vectors a, and a, denote t1-;:directlons of two-dim~gsional vectors A and B that make angles u and ,6, respectlvely. wltb a refdrence u-am, as shown ip Fig. 2-27. Obtaln a formula for the expans~on of the cosine of thp difference of two angles, cos(r - j), by takin, 0 the scalar product a , . a,,.

Fig. 2-17 Graph for 1 ' 0 Pfbblem P.2-5.

' ,

! P.2-6 Prove the law of sines for a triangle.

I :

P.2-7 Prove that an angle inpcgbed in a ~emicircle is a rigit in&. . i '. ..

P.2-8 Verify the back-cab rule of the vepor triple product of hree vectors, as expressed in Eq. (2-20) in Cartesian coordinates. ' : 1

1 ,) - '

P.2-9 An unknown vector can be dcter&ed if both its s&l& product and its vector product . w ~ t h 3 known vector are given, Assuming ;G is a known vector, determine the unknown vector S if both p and P are glven, where p = A X and P = A x X

>

P.2-10 Find the component of the vectqnA = -a,= + iz;'af the point P,(O, -2, 3), whlch is. directed toward the point ~ , ( j j . -60", 1).

I : i 1

PROBLEMS 63

P.2-11 The position of a point in cylindrical coordinates is specified by (4 ,2n/3,3) . What is the location of the point

a ) in Cartesian coordinates? b ) in spherical coordinates?

P.2- 12 A field is expressed In spherical coordinatcs by E = a,(25/R2).

a) Find [El and Ex at the point P(- 3,4, - 5). b) Find the angle which E makes with the vector B = a,2 - a,2 + a=.

P.2-13 Express the base :,-ctors a,, a,, and a, of a spherical coordinate system In Carresian coordinates. ,

P.2-14 Givcn a vcctor function E = a x y 3 a,x, evaluate the scalar line mtegral E - c lP from , PL(2, 1, - 1) 10 P2(8, 2, - 1)

a) along the parabola s = 2y2, b) along the straight line joining the two points.

Is this E a conservative field?

P.2-15 For the E of Problem P.2-14, evaluate J' E di from P3(3, 3, - 1) to P,(-l, -3, - i ) by converting both E and the positions ol'l', and P, into cylindrical coordinatcs.

P.2-16 Given a scalar function

a) the ~nagnitudc and the dircction of the maximum rate of increase of V at the point P(1 , 2, 3),

b) the rate of increase of Vat P in the direction of the origin.

P.2-17 Evaluate

(a,3 sin 8) - ds

over the surface of a sphere of a radius 5 centered at the origin

P.2-18 For a scalar function f and a vector function A, prove

V.(fA)= f V - A + A . V f in Cartesian coordinates.

P.2-19 For vector function A = a,r2 + a,2z, verify the divergence theorem for the circular cylindrical region enclosed by r = 5, z = 0, and z = 4.

-1

P.2-21 vector field I) = a,(cos2$)/KJ cxisls in the region bctwcen two spherical shells defined by R = 1 and K = 2. Evaluate

a) $ D . ds b) J'V-Ddv

P.2-22 A radial vector field is represented by F = aR/(R)i What do we know about the function f ( R ) i f V - F = 0 ? i

P.2-23 For two differentiable vector functions A and H, prove

V o ( A x H) = H - ( V x A) - E - ( V x A).

P.2-24 Assume the vector functiop A = a , 3 ~ ' ~ ' - ap3y2.

a) Find 8 A . d t around the triangular contour shown in Fig. 2-28. 6 ) Evaluate (V . A) - ds over the triangular area. C) Can A be,expmsed as the gradient of a scalar? Explain.

1 / I + Fig. 2-28 Graph for 0 2

x Problem P.2-24.

P.2-25 Gwen the rector function A = a, $in (4/2), verify Stokes's theorem over the hemispherical surface and its clrcular contour that are shown in Fig. 2-29.

1

I

, Grauh for.

x Problem p.2-2;

' P.2-26 For a scalar function f and a vector function G, prove

v x [ f G ) r f v x G + ( v f ) . x G

in Cartesian coordinates. . ,

P.2-27 Verify the null identities I

I ' a) V x ( V V ) r O b) V.(V x A ) s O j

i by expansion in general orthogonal curvilinear coordinates.

I

3 / Static Electric Fields

3-1 IN I HU~UCTION

I n Section 1-2 wc mcntioncct th;tt three essential steps x c involvcd in cc, nstructing a deductive theory for the study of a scientific subject. They are the definition of basic quantities, the development of rules of operation, and the postulation of fundamental .--., rclntions. Wc have cic(inci1 the S O I I S C ~ ;uid field ql~;~nfitics for 1I1c cIcct~-ol~~;~g~?t:tic' ~noticl in Chapter 1 and dcvclopcd the fundamenlals of vcctor algebra and vector calculus in Chapter 2. We are now ready to introduce the fundamental postulates for the study of source-field relationships in electrostatics. In electrostatics, electric charpcs (the sourccs) are at rcst. :~nd clcctric fields do ! k t cliangc with time. Thcre are I I O I I I : I ~ I I C I ~ C liclcls; I K I I ~ C W G cIc:11 will1 :I 1t1~1tivcIy S I I I I ~ I ~ sit~1:1tio11, Al'icr wc have studied the behavior of static clectric fields and mastered the techniques for sol-. ing clcutrnstatic boi~nclury-valuc problems, wc will then go on to thc subjcct of mayxtic fields and time-varying electromagnetic fields.

The development of electrostatics in elementary physics usually be, .ins with the experimental Coulomb's law (formulated in 1785) for the force between two point charges. This law states that the force between two charged bodies, q , and q 2 , that are very small compared with the distance of separation, R l 2 , is proportional to the . product of the charges and invcrscly proportion:~l to thc quar t : of the dist:incc, thc d i rcc t io~~ of the force being along the line connecting the charges. In addition, Cou- lonib found tha t unlike chargcs attract and likc ch;~rgcs rcpcl cach other. Using vector- notation, Coulomb's law can be written mathematically as

q1Cl2 Fl2 = -T' ( 3 - 1)

1- R12 where F12 is the vector force esertcdby (1, on q,, a,:,, is n unit vcctor in the direction from (11 to (I,, and It is :I l.rrol.rc\rtio~l;~liIy constant cicpcnding on the medium and the systcnl ol' units. Nok t h i ~ l i l 'q, and r12 arc of the same sign (both positive or both ncgalivc), I ; , , is positive (repulsive); and if q , and q2 are of opposite signs, F,? is negative (attractive). Electrostatics can proceed from Coulomb's law to define electric field intensity E, electric scalar potential, V, and electric flux density, D, and then lead to Gauss's law and other relations. This approach has been accepted as "logical,"

66 STATIC ELECTRIC FIELDS 1 3 , 1 2 ,

! #

perhaps because it begins with an pxperimental laGabsered in a laboratory and not with some abstract postulates. . J

We maintain, however, that Coulomb's law, though based on experimental evidence, is in fact also a postulate. Consider the two'stipulations of Coulomb's law: that the chargcd bodies be very small ~ompared wit9 the distance of separation and that the force is inversely propottianal to the squari: of the distance. The question arises regarding the first stipulatjonl,How small must! the charged bodies be in order to be considered "very small" dompared to the distahce? In practice the charged bodies cannot be of vanishing sizes (idcal point charges), and there 'is dificuity in determining the "true" distancz between two bodies ol finite dimensions. For given body sizes. the relative accuracy jn distance measurements is better when the separa- tlon is larger. However, practical cdnsiderations (weakness of force, existence of extraneous charged bodies, etc.) restrict the usable <istnpce of scparatrnn in the laboratory. and cxperlmcntal ~ I ~ ~ \ C C U S , \ C ~ C S canno[ hc cn~~rc ly :~vo~tlctl. '1'111s Ic,~tl$ LO

a more important question concenung the inverse-square relation of the second stipulation. Even if the charged bodies are of vanishing sizes, exp'erimental measurements cannot be of infin~te accuracy, no matter how skillfpl and careful an expen- mentor is. How then was it possible for Coulomb to know that the force was e x n c t l ~ inversely proportional to the square (not the 2.000001th or the 1.999999th power) of the distance of separation? This quesfhon cannot be answered from an experimental v~ewpoint because it is not likely that durlng Coulo~rrb's time cxperiments could have been accurate to the seventh place.+ We must thei-efore conclude that Coulomb's law is itself a postulate and that [he exact relation stipulated by Eq, (3-1) is a law of nature discovered and assumed by coulomb oq the basis of his experiments of limited accuracy. a , 1 ' '

Instead of following the hisjoripal development ,of electrostatics, we introduce the subject by postulating bothUibe d&?rgence and thje curl of th t electric field intensity in free space. From Helmholfz's theorem in ~ec t i bn 2-b 1 we kcow that a vector field 1s determined ~f its divergenqe and curl are specified. We deiive Gauss's law and Coulomb's law from the divergepce afid curl relatiobs, and do not present them as separatc postulates. Thc conccptl o f sc$nr potcnl i i~l fi)llr)wfi nat~~ral ly fro111 A VCL~OI.

identity. Field behaviors 'in material media will be studied and expressions for electrostatic energy and forces will be developed.

! I- I '

3-2 FUNDAMENTAL POSTULATES QF : . I \ ELECTROSTATICS IN FREE SPACE 1 .

We start the study of electroma&et$fn with the co&ideration of electric fields due to stationary (static) electric charges it7, free space. ~)ect'i-ostatics in free space is the

', ' " , < , s

' The exponent on the distance in ~ o u l o ~ b ' s ' l a w has been verifjdd by an lnd~rect experment to be 2 to w t h m one part In 10" (See E. R. Wilhqfls, J.$. Faller, and H : A ~ I H ~ I ~ , ph)~.~. Rev. Lerter~, vol. 26, 1971, p. 721.)

0 i n C- c i r r

U n CO

thi

ory and

.imental b's law: ion and juestion :n' order charged a l t y in ~r given scpara-

r' ,Ids due c is .

o be 2 to 26, 1971,

3-2 I FUNC-\MENTAL POSTULATES OF ELECTROSTATICS IN FREE SPACE

- C simplest special case OF electromagnctics. We need only consider one of the four fundamental vector field quantities of the electromagnetic model discussed in Section 1-2, namely, the electric field intensity, E. Furthermore, only the permittivity of free space 60, of the three universal constants mentioned in Section 1-3 enters into our formulation.

Electricfield intensity is defined as the force per unit charge that a very small stationary test charge experiences when it is placed in a region where an electric field exists. That is,

F E = lim - (V/m).

q-0 q

, The electricfield iatensity F i., ;r,,mg~on;il to and in the dircction of tllc force \ F. I f F is n~c;~sitrod in ~icwtons ( N ) ;ind cll;~lga '1 ill c o ~ ~ b i n b s (C), ~ilco C I 1s i n ~iewlons

per coulomb (NIC), which is the same as volts per meter (V/m). The test charge q, of course, cannot be zero in practice; as a matter of bct, it cannot be less tlian the charge on an electron. However, the finiteness of the test charge ivould not maks the measured E differ appreciably from its calculated value if the test charge is small enough not to disturb the charge distribution of the source. An inversc rc~atibn of Eq. (3-2) gives the force, F, on a stationary charge q in an electric field E:

The two l ~ ~ d : l i n ~ ~ i t i t l postulates of clcctros~atics in lice sp:tcc spccify ttic divergence and curl of E. They are

In Eq. (3-41, p is the volume charge density (C/m3), and so is the permittivity of free space, a universal constant.' Equation (3-5) asserts that stoiic elecrricjrlds are - irrotntB~niTt-wl~erc;~s Eq. (3-4) implics ~ l i ; ~ t :1 st;~tic electric ficlcl ir n& roIcnoid:~l unless p = 0. Thasc iwo postulales arc concise, simple, and indepenhent of any coordinate systcm; and they can be used to derive all other relations. laws, and I ~ I T I W I B r l w l ~ u s t i l l i c s I S u d ~ is lhe I m t ~ i y of ills dtduc~ivc , :~xio~~l;llic :~ppm:dl.

1 The permittivity of free space r ; z - x (F/m). See Eq. (1-11)

36n

68 STATIC ELECTRIC FIELDS I I C*

7 I ' I i

f j:. I 8 Equations (3-4) and (3-5) arq polnt relations; that is, tbeyhold at every point

in space. They are referred to as thq diffeiential form of fhe pqstulates of electrostatics, since both divergence and curl operations involve spatial derivatives. In practical applications we are usually interested in the total field ofan aggregate or a distribution of charges. This is more conveniently 'obtained by a;? integral form of Ey. (3 -4). Taking the volume integral of both sides of Eq. (3-41 over,*an arbitrary volume V, we have ! c 2 I

" , 1 1 "

J v P- E d r = 2. €0 J v dv.

In view of the divergence theorerq in Eq. (2-104), Eq. (3 -6) becomes

where Q is the total charge contained'in volume V bounded bi;ii7face S. Equa- tion (3-7) is a form of Gauss's luw, which statcs thaiithe ~ptul ourwardjlux of' the electricfield intensity over an): closed surface in fi-ee spice is equal to the total charge enclosed in the surfilce divided hy E,,. Gauss's law is one i f the most important relations in cicctrostatics. Wc will discuss i t furthcr in Scctiol! 3 - 4, along with illustrative examples.

An integral form can also be 06tained for the hurl relation in Eq. (3-5) by integrating V x E over an open : I . surfack , . and invokingiStokes's theorem as espressed in Eq. (2-131). We have . .

%. 1 : ) -

The line integral is performed over a closed contour C-bounding an arbitrary surface; hence C is itself arb~trary. As a yatte? of fact, the surface does not even cntcr into Eq. (3-8), which asserts that the . y u l ~ r line inlegrul oJ',rhe stutic clcclric Jicld rtttcrl.sily arotuzd an!. closed path vanishes. ~ h i s i s s i m ~ l ~ anothef way of saying that E is irrotational or conscrvative. Refcrrinito F&. . . 3-1. wc scc;,th:lt i f the scalar linc intcgrnl

I

P

3-3 COU

-ch; r ?a, . ...-

Fig

6 ry point .k: 3. ostatics, I $

~ractical :?

' .-G

ribution , ji 4. (3-4).

8 , lume V,

;i-, t;

3-3 / COULOMB'S LAW 69 ,

of E over the arbitrary closed contour C,C, is zero, then

JPp,' E . d P = -J: E . & (3-10) Along C, Along C,

or

SpT E.dP=SP:' E - d P Along C, Along C,

, Eq~~a t ion (3-1 1.1 says that the sc:ilar .linq i t~lc~ra!- or the i r ro~ :~ t i&~l ;~ l F lic!,l I, I I I , , ~

I X I I C ~ C I I ~ tlllhe I X I ~ I I : i l ~ C ~ C I I & i ~ ~ l y 0 1 1 ~ I I C C I I C I l > ~ i l ~ l s . i\s \"I. sI1:111 scc i l l SCC~ioll .i 5. tile integral in Eq. (3-11) represents the work done by the electric field in moving a unit charge from point P , to point I',; hence Eqs. ( 3 4 ) ;ind ( 3 9 ) imply a st:itcmcnt of conservation of work or energy in an electrostatic iield.

The two fundamental postulates of electrostatics in free space are repeated belarv because they form thc h~tnd:~t ion ~ I P O I I \ V I I I C I I wc build I I I C s ~ r ~ i ~ i i ~ i c ~ l c l ~ ~ ~ r o s ~ : ; ~ i i s .

3-3 COULOMB'S LAW

We consider the simplest possible electrostat~c problem of a s~ngle polnt charge, y. at rest in a boundless free space. In order to find the electr~c ficld intanslty due to (1. we draw a hypothetical spherical surface of a radius R centered at q. Since a polnt charge has no preferred directions, its electric field must be everywhere radial 2nd - bas the si1111c intansity : ~ t :ill poiots on 1111: ~ l ~ l i ~ l l . i ~ i l l S U I ~ I C C , Apply~ng Eg. (3-7) to I?g, 3 - 2(:r), we l ~ v c

70 STATIC ELECTRIC FIELDS I3

(a) Point charge at the origin, (b) Point charge not at thc origin.

Therefore,

L

Equation (3-12) tells us that ille electric jirld intensity o j u point charge is i s the ourward radial direction and bqs a magairude propoytional ro the charge and incersely proportional to the square of the distance born the charge. This is a very important basic formula in electrostatics. It i s readily verified:tllat V x E = 0 for the E given in Eq. (3-12).

t

I l the chitrgc q I \ not locutcd at the orlgln of u choseq cuord~naio \ystcnl, 5i11iablc changes should be made to the unit vector a, and thb distapce R to reflect the locat~ons of the charge and of the point at which E is to be dbtermieed. Let thc position vector of q be R' and that of a field poiht P be R, as shown in Fig. 3-2(b). Then, from Eq. (3-1 2), )

where a,, is the unit vector drawn'fiorn q to P, Since

we have

Example 3-1 Determine the electric fieid intensity a1 P(-0.2,0, -2.3) due to a point charge of + 5 (nC) at Q(0.2, $1, - 2.5) in air. All dimensions are in meters.

I 1 .

3-3 I COULOMB'S LAW 71

Solution: The position vector for the field point P

The position. vector for the point charge Q is

R' = = a,0.2 + ayO.l - a,2.5

The difference is

which has a magnitude

. jR - 2'1 = [(-u.4)' + (-0.1)' + (0.2)3]112 = 0.455

Substituting in Eq. (3-15), we obtain

'I'hc qwtlticy w i ~ l l i ~ ~ l l ~ c ~ ~ : ~ ~ . c ~ i ~ l ~ c s c s is thc' [Init vcctor :lu,. -= (I1 - l1')jIR -- R ' / . ; I I I ~ I(:,, 11:ts :I 111:1g1i~tdc CII' 214.5 (V/III).

Note: The permittivity of air is essentially the same as that of the free space. The factor 1/(4ne,) appears very frequently in electrostatics. From Eq. (1-1 1) we know that c0 = l / ( ~ ~ p ~ ) . . B ~ t pO = 4n x lo-' (H/m) in SI units; so

exactly. Ifwe use the approximate value c= 3 x 10"nl/s), then l / (bc , ) = 9 x 109 (m!F).

When a point charge y, is placed in the ficld of another point chargc (1, 'lt the origin, a force F,, is experienccd by 4, due to electric field intensity E,, of q , at y2. Combining Eqs. (3-3) and (3-12), we have

Equation(3-17) is a mathcmatical form of Coulomb's luw alrcady statcd in Scction 3- 1 in conjunction with Eq. (3-1). Note that the exponent on R is exactly 2, which is a consequence of the fundamental postulate Eq. (3-4). In SI units the proportionality constant k equals 1/(4nc,), and the force is in newtons (N).

I

72 STATIC ELECTRIC FIELDS / 3

L $ 1

-" + Screen \I :

Deflection -f

P ,, I

Fig. 3-3 Electrostatic deflection system of a cathode-ray O \ C I ~ ~ ~ I ~ K I ~ ~ ( ~ \ ~ l l ~ l p ~ ~ ? 2 )

Example 3-2 The electrostatjc de1:cclion system of a cathode-my oscillograph is depicted in Fig. 3-3. Electrons from a heated cathode are given an initial velocity vo = a , ~ , by a positively charged anode (not shown). The electrons enter at r = 0 into a region of deflection plates where a uniform electric field Ed = - a,Ed is maintained over a width w. Ignoring gravitational effects, find the vertical deflection of the electrons on the fluorescent screen at r = L. i

Solutiot~: Since there is no force in the ;-direction in the z > 0 region, thc horizontal velocity v0 is maintained. The field .Ed exerts a force on the electrons each carryng a charge - e, causing a deflection in the y direction.

From Newton's second law of motion in the vertical direction, we have

where rn is the mass of an electron.~1~tegrating hoth.sidcy, wc obtain

where the constant of integration isset to zero beCause v,. = 0 at t = 0. Integrating again, we have

The constant of integration is again zero because y = 0 at t = 0. Note that the electrons have a parabolic trajectpry between the aeflection plates. At the exit from the deflection plafes; t = w/vo,

'

3-3 I COULOMB'S LAW 73

' and

' When the electrons reach the screen they have traveled a further horizontal distance of (L - w ) which takes (L - w)/v, seconds. During that time there is an additional vertical deflection

Hence the deflection at the screen is

3-3.1 Electric Field due to a System of Discrete Charges

Supposc a n clac1rost:liic liald is cra:ilcd by a group 01' i i dlscrete po~ot chiirgel ' I , , q2 , . . . , q, located at different positions. Since electric field intensity is a linear func .on of (proportional to) a,q/~', the principle of superposition applies, and the tot;.] E field at a point is the vector sum of the fields caused by all the individual char: CS. From Eq. (3 -15) wc can wriic l l ~c clcctric ilitcnsity at a field point whose posir on vcctor is It as

Although Eq. (3-18) is a succinct expression, it is somewhat inconvenient to use, because of the need to add vectors of different magnitudes and directions.

Let us consider the simple case of an electric dipole that consists of a pair of equal and opposite charges, + q and -q , separated by a small distance, d, as showc in Fig. 3-4. Let the center of the dipole coincide with the origin of a spherical coordicate System. Then the E field at the point P is the sum of the contributions due to +q

Fig. 3-4 dipole.

Electric field of a

74 STATIC ELECTRIC FIELDS 1 3 j . L 1 .

* 8

. ; t : and - y. Thus. + %

The first term on the right side of ~4.:(3-19) can be Simplified if d << R. We write

where the binomial expansion has been used and dl terms containing thc second and higher powers of (d/R) have beeri neglected. Similarly, for the second term on the right side of Eq. (3-19), we have'

Substitution of Eqs. (3-20) and (3-21) in Eq. (3-19) leads 10 I

The derivation and intcrprolati.b$ of Ey. 13--22) rcqrirc thc manipulation of vcctor quantities. Wc can apprcciatcLhat tictcrrninihg thc clcctric Geld causcd by three or more discrete charges will 'bg even morc t&liouq; I n Scction 3 .5 wc will introduce thc conccpt or ;I scalt~r clcctric potcl~li:~l, will1 whic l~ [ l ~ c clcclric licld intensity caused by a diswibution of'cbargcs can be found more easily.

The electric dipole is an irqporkant entity in tGe study of the electric field in dielectric media. We define the product of the charge q an4 the vector d (going from - q and +q) as the electric dipoff mement, p:

, * . - -p = yd . : .

i (3-23)

Equation (3-22) can then be rewritten as > .

where the approximate sign ( - ) ovd; 'the equal sign 'has been left out for simpliaty.

! I

8 I ) '

(3-19)

2 write

3-3 1 COULOMB'S LAW 75 '

If the dipole lies along the z-axis as in Fig. 3-4, then (see Eq. 2-77)

p = aZp = p(a, cos 8 - a, sin 8) - - .. (3-25) '

R . p = R p cos 0 , (3 -76)

and Eq. (3-24) becomes

E = - ( a , 2 cos 8 + a, sin 6) (V/rn). 4m0R (3 -27)

Equation (3127) gives the electric field intensity of an electric dipole in sphcri~n! . coordinates. We see that E of a dipole is inversely propoirional to the cube of the distance R. This is reasonable because as R increases, the fields due to the closely spaced + q and - q tend to cancel each other more completely, thus decreasing more rapidly than that of a single point charge.

7

3-3.2 Electric Field due to n Continuous Dlstrlbution of Charge

The electric field caused by a continuous distribution of charge can be obtained by integrating i5kpcrposing) thc contribution of a n element of c h a r p over the charge clislrihulio~l. licfcr to 1 . i ~ . 3 -5. whurc ; I voluinc cL;irp: tlisfrih~itioi~ is s l ~ o w i ~ . T ~ I C V O I U I I I ~ ~ * C ,~ I : I I ,K( : I ~ ~ - I I . I I I ~ ,I I ( / I I I I ) 1-1 :I I I I I I L I i 0 1 1 0 1 ~ I I C C O O I ~ ~ I I I : I ~ C : . . S I I ~ L C :I iIilli~c1111:11 clement ul' charge behaves like a point charge, the contribution of the charge p -10' in a differential volume element du' to the electric field intensity at the field point P is

We have

p dv' dE = a, -----.

4n.5,R2

. Fig. 3-5 Electric field due to a continuous charge distribution.

h.

. I I i. > I i . '

76 STATIC ELECTRIC FIELDS I 7 I

I

! ! I *

or, since a, = R/R, . , , 3

(3-30)

(

Except for some especially sikp14 ases , the vectdr triple ibtegral in Eq. (3-29) or I

Eq. (3-30) is difficult to carry out because, in general, all'three quantities in the integrand (aR, p, and R) change with the location of the differential volume dv'.

If the charge is distributed on a,surface with a ~ u r f a c e charge density p, (C/m2), then the integration is to be carried out over the surface (not necessarily flat). Thus, 8

For a line charge, we have 1..

where p, (C/m) is the line charge 'density. and L' the line (not necessarily straight) along which the charge is distributed.

I

Example 3-3 Determine the electric field inten& of Ian infinitely long. straight. line charge of a uniform density p h air. i

4 ' -

Solution: Let us assume that tlje.line charge l i & alopg the zf-axis as shown in Fig. 3-6. (We are perfectly frce to dd this because t\le field obviously docs not depend

. . on how wc clcsign;~tc thc linc; 11 ?,> ' i r rr u r w p / ( * d cvrricir~ri/i,;n f r r r i : i ~ pr i tn!d r~r~orrlitrrc~c~.v

jbr sotircc / ~ r , i t ~ / s wid ~ ~ t ~ ~ ~ r i ~ ~ ~ ~ ~ ~ ~ ~ c r ~ r ~ ~ ~ ~ ~ l ~ r ~ ~ i / ~ ~ , s / ; , I * , / I v / ~ / / ) ! , ~ I I / . S ~ W I I 111m~ i s ( 1 / r r r . $ . $ ~ / ~ ~ l i / y of conjusion.) The problem agks us to find the e ~ & ~ i c geld intensity at a point l', which is at a distance r from the I'ine. Since the p+blern has a cylindrical symmetry (that is, the electric field is in!dep&dent of the azlmutli angle 4). i t woiild bc most convenient to work with cylijdriyd coordinates. h e re writ;.^^. (3-32) as . - . . .

For the problem at hand p, is co&&nt and 5 line element dLr = dz' is chosen to be at an arbitrary distance z' from the,Grigin. It is mqit important to remember that R is the d~stance vector directed fr;h the source td'thejeld point, not the other way

I*

: i . . ii

! ' : *

A 1: ,.i

1 i : i l

I

3-3 1 COULOMB'S LAW 77

i - 29) or s in the iL;', , [Cim2), t). Thus,

Fig. 3-6 An infinitely long straight-line charge.

around. We have R = a,, - a,:'.

The electric field, dE, due to the difirential line charge element p , d t ' = p , d l is

where

and

p,r dz' dEr =

4 m 0 ( r 2 + 2'2)3/2

- p,zf dz' dE, = 4neO(r2 + z '2)3/2 '

In Eq. (3-35) we have decomposed d E into its components in the a, and a; directicns. It is easy to see that for every p, dz' at + z' there is a charge element p, d:' at - :', which will produce a dE with components dEr and -dE,. Hence the a= components will cancel in the integration process, and we only need to integrate the dBr in Eq. ( 3 -35a):

78 STATIC ELECTRIC FIELDS 1 3 ,

Equation (3-36) is an impor t a~~ l i.csult for an inhiite linc chargc. Of course, no physical line charge is infinitely long; ncvcrtl~cless, Eq. (3-36) gives the approximate E field of a long straight.line charge at a point close to the line charge.

3-4 GAUSS'S LAW AND APPLICATIONS !

Gauss's law follows directly from the divergence p&tulate of elect~ostatics, Eq. (3-4), by the application of the divergence theorem, 1t;has been derived in Section 3-2 as Eq. (3-7) and is repeated here on account of its impoytance:'

I --.----.--.---,- 1

Gtruss's Irr\\. osscrrs rlrtrr rlrc rortrl ool\\~rr~tl,/lrr.\- (!I ' tlrc~ IG/icItl o r w t r ~ ~ j - c~/o,scd s~rr:/ircv in p e e sptrce is o q 1 ~ 1 1 to the ruttrl C . / I ~ I I Y I C ~ C I I ~ I O S C Y I ill tire S L I I $ I C ~ ~lil:ided by E ~ . We note that the surface S can be any hj~pothetical (nzathrr~atical) close&rJace chosen for come?lierlce: it does not have to be. and usually is not, a physical surface.

Gauss's law is particularly usefui in determining he E-field of charge distributions with some symmetry conditions, such that the noynal c&lporreitt *f tlze electric ,field irllcrlsil!~ is c w l s ~ r r ~ i ~ oocr (111 cv~c~lo,sc~l .srr~;luc.c. I n such cxcs tllc surfr~cc intcgral on thc left side of Eq. (3-37) would bc very easy to cval.u;~te, &nd Gauss's law would be a much more efficient way for fin'ding the electrik field intensity than Eqs. (3-29) through (3--33). On the ather hapd. whcn synimeti-y conditions do not exist. Gauss's law would not be of much help. TL essence of applying Gauss's law lies first in the recognition of symmetry conditions, and second in thqsdtable choice of a surface over which the normal component of E resulting fro? a given charge distribution is a constant. Such n surface i s referred to as a Ga~~ssiar? su r f kc . This hasic principle wiih L I S ~ t o o1mi11 l<q. ( 3 131 1'01; ;I, poi111 ~ I ; I I ~ ; , c ~ I I ; I I p o s ~ c s s ~ s spl~c~.ical < , Y I I I I I I C I ~ Y ;

consccl~~cnlly, a propcr C;a~~ssi:~lr +tirliic,: i:., ~ l i c < L I I . I : L ~ : C ! O ~ ' ;L ~ ~ [ ~ I I ( : I ~ G C C I I I I : I ~ : I I L ~ I C

point charge. Gauss's law cc)uld nbt help in the Ycrivqtion of Eq. (3-22) or (3--27) for an electric dipole; since a surfice about a sephrateb pair of equal and opposite charges over which the norqal coinponent of E-rzmains constant was not known.

1 i . t

P Example 3-4 Use Gauss's law tobetermine the electric field intensity ofan infinitely long. straight, line charge of a uniform density in air.

Solution: This problem was solved in Example 3-3 by using Eq. (3-32). Since the line charge is infinitely long, the resultant E field must be radial and perpendicular to the line charge (E = a,E,), aqd a component of E along the line cannot exist. With the obvious cylindrical symmetry, we construct q'cylindrical Gaussian surface of a radius r and an arbitrary lenith L with the line charge as its axis, as shown in Fig. 3-7. On t h ~ s surface, E, is cdnstant, and ds = a,r d 4 dz (from Eq. 2-52a). We

I i 1

3-4 1 GAUSS'S LAW AND APPLICATIONS 79

have

Infinitely long . . uniform ling charge, pp. Fig. 3-7 Applying Gauss's

law to an infinitely long line charge (Example 3-4).

There is no contribution from the top or the bottom-face of the cylinder because on Ihc top f x c 11s ---- :I2r tlr r l $ but li, has no z-component there, making E ils = 0. S ~ I I I ~ I : I I I Y I'or Ilic I I I I I ~ O I I I I ~ I ~ G , ' l l ~ ~ O I J I I ~ I I I I . ~ C C I ~ ~ : I C ~ + C L I i ~ i ~ I I C C ~ I ~ I I C I C I ~ is Q -. ,,,<L, Substitution into Eq. (3-37) gives us immediately

This result is, of course, the same as that given in Eq. (3-36). but it is obtained here in a much simpler way. We note that the length, L, of the cylindrical Guassian surface does not appear in the final expression; hence we could have chosen a cylinder of a unit length.

Example 3-5 Determine the electric field intensity of an infinite planar charge with I. a uniform surface charge density / I , .

S n l ~ l f i o ~ l : I t i s clc:t~. tllitl 1 1 1 ~ ' E licld ci1115ctI hy a clurgctl sltccl ol'an inlinite extent is ~lurmal Lo llic shccl. Equalion (3-31) could bc used to find E, but this would inv~lve a double integration between infinite 1irn;ts of a general expression of 1/R2. Gauss's law can be used to much advantage here.

I

3 - 1 ' . 80 STATIC ELECTRIC FIELDS / 3 ' : w

: f I r 1 ' I ,

5 :

surface charge, p, 8 '

Fig. 3-8 : Applying Gauss's law to anhfini te planar charge (Example 3-5). ,

We choose 3s the Gaussian ourSac,c n rectangular box with top and bottoln t iccs of an arbitrary area A equidistant from the planar charge, as shown in Fig 3-5. The sides of the box are perpendicular to the charged sheet. If the charged sheet coincides with the xy-plane, then on the top face.

- . . E . ds = (a,E,) .(a,ds) = 6,d.s.

On the bottom face,

E - ds = (-a,E,). ( -a , ds) = E,ds .

Since there is no contribution from the side faces, we have

6 E ds '= 2Ez ds = 2 2 , ~ .

The total charge enclosed in the b o i i's Q = p,A. heref fore^

from which we obtain I

and

' ,

Of course, the charged sheet mqy j o t coincide with the v-plane (in which case we do not speak in terms of above 'dd below thk plane). but the E field always points away from the sheet if ps is positive. s

3-4 1 GAUSS'S LAW AND APPLICATIONS 81 \

Example 3-6 Determine the E field caused by a spherical cloud of electrons with . a volume charge density p = -p, for 0 P R P b (both p, and b are positive) and , p = O f o r R > b .

Solution: First we recognize that the given source condition has spherical symmetry. The proper Gaussian surfaces must therefore be concentric spherical surhces. We must find the E field in two regions. Refer to Fig. 3-9.

A hypothetical spherical Gaussian surface S, with R < h is constructed with111 rhc electron cloud. On this surface, E is radial and has a constant magnitude.

E = a;,E,, . - rls = a, d s . The total outward E flux is

6, E - ds = ER Js dr = ER4nR2.

The total charge enclosed within the Gaussian surface is

e = J, p

I (Example 3-6).

I . Substitution into Eq. (3-7) yields i

f

We see that within the uniformelectron cloud-the J3 field is directed toward the center and has a magnitude pfdportiona~ to the distance from the center. . . , ,

b) R 2 b I , \.I P

For this case we construpt a fpherical Gaussiaq &face So with R > h outside the electron clouil. We obtain ihe same expression lor jL0 E ds as in case (a). The total chargc e~closed is , h

' .

which follows the inverse qquare law and could have been obtained directly from Eq. (3-12). We observe that aurside the charged claud the E field is exactly the same as though the total chafige is concentrated on a single point charge at the center. T h ~ s is true, in general, for a spherically symmetrical charged rcglon even though p is a function of R.

The variation of ER versus R is plotted in Fig. 3-9. Note that the formal solution of this problem requires only a fe% lines. If ~ a u s s s law is not used. it is necessary (1) to choose a differential volume dement arbitr&ily located in the electron cloud, (2) to express its vector distance R: to a field poi4t in a chosen coordinate,syitem, and (3) t6 perform a triple integration as indicated'in Eq, (3-29). This ir a hopelevly involved process. The moral is: T!) to apply Gdhssls.law if symmetry conditions exist for the given charge distributicn.

, t h i: ,

. a

3-5 ELECTRIC POTENTIAL: I .. <

In connection with the null identhy in Eq. (2-134 we poted that a curl-free vector field could always be expressed ,zs the gradient of'a sc, jar field. This induccs us to 1 '

define a scalar electric potentiql, c, such that , . ! '

(3- 38)

, because scalar quantities are aasidr ;to handle thad vect r uantities. If we can deter- 5 mine V more easily, then E cap b& fbund by'a gradient operation, which is a straightforward process in an orthogonal~coordinate system. The reason lor the inclusion of a negative sign in Eq. (3-38) $ill be explained pr&eerltly,

t a -

)In faces 'ig. 3-8. :d sheet

3-4 / GAUSS'S LAW AND APPLICATIONS 81 ,

Example 3-6 Determine the E field caused by a spherical cloud of electrons with a volume charge density p = -p, for 0 5 R i b (both po and b are positive) and '

, p = O f o r R > b .

Solution: First we recognize that the given source condition has spherical symmetry. The proper Gaussian surfaces must therefore be concentric spherical surfaces. We must find the E field in two regions. Refer to Fig. 3-9.

A hypothetical spherical Gaussian surface Si with R < h is constructed within ihs electron cloud. On this surface. E is radial and has a constant magnitude.

E = a,$,, . - r l s = a, d s .

The total outward E flux is

6, E . ds = E, Js ds = ER4nR2.

The total charge enclosed within the Gaussian surface is

e = j', p du

4n. = - I = - -

3

l;i&, 3 -0 lilcclric l ic ld Inlcnslty of' a sphcrical electron cloud (Example 3-6).

aard the I-.

1 outside case (a).

f-

11) ,..?rn L!LYl, c at the ion cvcn

\olutloll eccssary In cloud, 's) stem, ipclessly )nditions

:e vector 2es us to ,

r (3-

! I \ t l c l w ~trn~ght- luslon of

3-5 1 ELECTRIC POTENTIAL 83

Electric potential does have physical significance, and it is related to the work done in carrying a charge from one point to another. In Section 3-2 we defined rhe

, electric field intensity as the force acting on a unit test charge. Therefore, in moving a unit charge from point P , to point P , in an electric field, work must be done against the,jield and is equal to '

'Many paths may be followed in going from P , to P , . Two such paths are drawn in ~ i g . 3-10. Since the ~ a t h between P , and P, is not specified in Eq. (3-39). the question naturally arises, d o ~ i iilc work dcpund on the p t h i:lkcot! h liitlc il~ou$it i i i ; i lead Gs'to conclude fhat CV/q in Eq. (3-39) should not depend on the path; for. l f i t did, one would be able to go from P1 to P, along a path for which W is smaller and then to come back to PI along another path, achieving a net gain in work or energy. This would be contrary to the principle of conservation of energy. We have nlrcndy alluded to the path-independence nature of the scalar line integral of the irrotationni (conservative) E field when we discussed E q (3-8).

Analogous to the concept of potential energy in mechanics, E q (3-39) represents the difference in electric potential energy of a unit charge between point P, and point P I . Denoting the electric potential energy per unit charge by V, the electric potenrial, wc have

Mathematically, Eq. (3-40) can be obtained by substituting Eq. (3-38) in Eq. (3-39). Thus, in view of Eq. (2-81),

-Spy E . dP = SP'' ( V V ) . (al d l )

Fig. 3-10 Two paths Icading from P , to P, in an electric. field.

84 STATIC ELECTRIC FIELpS /:J' ! , .

: 4 1

Direction of iwrcming V '

Fig. 3-1 1 Rclutivc direction\ u of and increasing V.

. ; j ,. . , ,

What we have defined in Eq, (3-40) is a ptent id difference (elcrrrorroric i.oltoge) between points P , and P I . It makes no moresense l;o talk about the absolute potential of a point than about the absolpte phasc of a, phasor ar thc absolutc altitude of a geographical location: a refercpceqro-potential point. a reference zero phase (usmlly :I( r :; 0). or :I rclixc~~cc zc1.11 ;~lli!h!c ( I ISI I : I I I~ ; I ( S Y I w u I ) I U \ I S ~ lirst hb sl)chyiliccl. 111

most ( 1 ~ 1 t not i l l ) G I S ~ S , tlic ~ c ~ ~ o - p i ~ l ~ ~ ~ ~ i ; ~ l l;o1111 i s l4Lc11 ;I! i11Ii11ity. W I I ~ I I (lie I V I ~ ~ L - I I C ~ zero-potential poinl is 1101 ; \ I inlini!y, i t shoulri bc-spcci/icully statctl.

We want to make two mqre about Eq;(3-38). First, the inclusion of the negative sign is necessary in ordoe to copform with tile conve&on that in going oyoiirst the E field the electric potenlial V ina.eo.ser. For instance. when a DC battery of a voltage Vo is connected between two parallel qondu~ting plates. as in Fig. 3-1 1, positive and negative charges;cu$ulate, respectively,: 011 the top and bottom plates. The E field is directed from pdsitive to negative chitrges. whiic thc potential incrcnscs in the opposilc direction. Sccqn?.'dx know from &&tiqp 2 --5 when we dclincd the gradient of a scalar field that the dirkction of V V isiiormyl to the surfaces of constant V . Hence, i f we use directed flili'(iljir~e.s or .strennili+s~to indicate the direction of the E field. they are everywhere perp&diculitr to qiiipotu@iul lines and ~qsiporoliid

. . : !' . S U I ~ ~ J C ~ ' S . ' I ! 1

3-5.1 Electric Potential due to a I .

Charge Distribution ! I I '

The electric potential of a at 2 distance R fr* a point charge q referred to that at infinity. can be obtained rGdily from Eq. (3-43:

, I - , d . .

which gives I

(3 - 42)

' I r , This is a scalar quantity and on, besid~sq,~only the distance R. The potential difference between gny two poiq ts '~ , and PI at bistalfes R, and R , , respectively.

, ? I ' . ! :

, I I . I . . i ; i t :

1 , ' a ! i .

! 4 ; I 8

3-5 / ELECTRIC POTENTIAL 85

f

,,---\ /'

/ /' ,/-- $'\

1 \ I

/ I \

1

I I \ 9 R~ I I \ I I

/ \ '\ /' I \ '---/ /

/ '\ I

$ / '\ / , ' Fig. 3-12 Path of integration '%---- about a point charge.

. $

from q is

' l l ~ i s rcsult m ~ y ;~ppc:~r :I little siirprisi~~g a1 first, si~lcc lJ2 ;~nd 1 1 ~ y no1 lie on tile same radial line through q, as illustrated in Fig. 3- 12. However, the concentric circles (spheres) passing through P, and P, are equipotential lines (surfaces) and Vp2 - 1., is tllc S : I I ~ :IS I<,, - V,.,, Fro111 I I I C poi111 d v i c i of l<q, ( 3 40) wc G I I I c l ~ ~ ~ o s c 111c p ; ~ t l ~

01' i ~ l l c p y ~ ~ i o ~ l I ' I ' O I I ~ l ' , 111 I * , i l l i d ~ I I C I I liu111 l b i 10 /Ii. N U W O ~ L is ~ I C ~ I I C I ~ U I I I 1 ' : 10

P , bccause E is perpendicular to dP = a,R, dq5 along the circular path (E - clC = 0). The electric potential due to a system of n discrete point charges q,, q,, . . . . qn

located at R'I, R;, . . . ,R: is, by superposition, the sum of the potentials due to the individual charges:

Since this is a scalar sum, it is, in general. easier to determine E by taking the negative gradient of V than from the vector sum in Eq. (3-18) directly.

As an example, let us again consider an electric dipole consisting of c h a ~ e s t q and - q with a small separation d. The distances from the charges to a field point P are designated R+ and R-, as shown in Fig. 3-13. The potential at P can be written down directly : '-

If d << R, we have

V P

1 !

5 I 11 I

71

S 1

1 ,

.Fig. 3-13 :An elq~trid dipole.

I

and 1 d ' - 1 (1

cos 0 . R_z(R+jms:O) Z R - ' ( I ; ~ ~ )

qd cos 8 Y!= --- -1

4ncOR2

I

where p = qd. (The "approxima!e" sign (-) has bee0 dropped for simplicity.) The E field can be obtaineq fro'rn - VV. In spherical coordinates we have

: L

= --. ( a , ~ cos o + a; sin 4). (3 491 47T601<3 - L~ a

Equation (3-49) is the shme as eq.~3-27), but has Feen bbtained by a simpler procedure without manipulating psit ti on vectors. I:

Example 3-7 Make a twq-d@en;ional sketch of the q~uipotential lines and the electric field lines for an electrii; dipole. : . T

I . ' Solution: The equation of an eqiipotential surfade of q charge distribution is obtained by setting the expressiop for V to equal a constant. Since q, d, and E, in Eq. (3-48) for an electric dipole afe fixe'd quantities, a constant V requires a constant ratio (cos 0/R2). Hence the equation' for an equipotential surface is

1

:, Pi= c , & a , ' - ' , (3 -50) 3 .

' t , 9

;: I

I. : . 1. , . i a

I . .. .

I

( 3 - 37)

/?

( 3 3 8 )

I \ f2

( 3 -19)

~ I c r pro-

and the

P\ m is ob- - 0 111

constant

(3 -50)

3-5 / ELECTRIC POTENTIAL 87 ,

F

where c, is a constant. By plotting R versus 8 for various values of c,, we draw the solid equipotential lines in Fig. 3-14. In the range 0 i 8 I n/2, V is positive; R is maximum at 8 = 0 and zero at 8 = 90". A mirror image is obtained in the range

' n/2 I 8 S n where V is negative. -4 3

The electric field lines or streamliiks represent the direction of the E field in space. We set -,

fig. 3-14 Equipotential andrlectric field lines of an electric dipole (Example 3-7).

. I

, i 1 :.,

8 * I

s J . 88 STATIC ELECTRl 1 . ,

b . i d 1

2:

where k is a constant. In spheri 1 &?ordinates, Eq.43-51) becomes (see Eq. 2-66). r , L

a , dR + aeR dB + aJ? pih8 d+ = k ( g R f 4 ~ *Ee + a$.&, (3-52) 5 . .

which can be written I

dR R.d8 R sin B d4 -=I- - (3-53)

. iE8 IE, Eq, '

For an electric dipole, there is na E4 :component, arld I i .. 'dR Rd8 " :

= - 2 cos 8 sin 0

or 46: 2 (/(sin a) -= 1.1 -54) p sin 0

Integrating Eq. (3-54). we obtajp : ; -1.

R h cE sinZ 8, (3-55)

where c, is a constant. The electriq field lines, having m a x p at 8 = n/2, are dashed in Fig. 3-14. They are rotqtion@lly sfmmetrical abdut thq z-axis (independent of 0) and are everywhere normal to !fie equipoteptial l$?s, u

1

The electric potential due to a ; c p h u o u s qistributiqn of charge confined in a I eiven'region is obtained by intdgrafirlg the contribpdon of an element of charge over

;he charged region. We have, fqr q ldlume charge &itrib tion, I 8

I 1 '"

For a surface charge distributi~g, I- ,

Example 3-8 Obtain a foqulti*For the electljo~field4intensity on the axis of a circular disk of radius b that clrrieb a uniform surface charge density p,.

, : I I t

!; . ' ;, ': i 1

i i j . !; 1, $

).

,i ! I , !, :

. 2-66).

(3-52)

# (3-53)

(3 -54)

(3-55) r

d a s i d n t c

led in a -ge over

(3 -56)

(3 -57)

t n

(3--""'

xis of a

' t 3-5 / ELECTRIC POTENTIAL 89 ,

3 I

Solution: Although the disk has circular symmetry, we cannot visualize a surface around it over which the normal component of E has a constant magnitude; hence Gauss's law is not useful for the solution of this problem. We use Eq. (3-57). Working

I with cylindrical coordinates indidated in Fig. 3-15, we have I

ds' = r' dr' d#' and

R = Jmi. The electric potential at the point P(0, 0, z) referring to the point at infinity is

r ' I

dr' d4' 1

Pn =- [ (z2 + b2)'I2 - 1211.

260 (3 -59)

Therefore, E = -VV= - av

a, - az

The determination of E field at an off-axis point would be a much more dificult problem. Do you know why?

For very large z, it is convenient to expand the second term in Eqs. (3-60a) and (3-60b) into a binomial series and neglect the second and all higher powers of the ratio (b2/z2). We have

Fig. g l 5 A uniformly charged disk (Example 3-8).

1 ' ' I . ; ; . , . , Substituting this into Eqs. (3-Pa) &d

I 4'

I E=a, - ~ K E ~ Z ~ .

where Q is the total chargp on ihe disk. Hence, when the point of observation is very far away from the charged disk, thpE field approximatelj follows the inverse square law as if the total charge were coqcentrated at a point. I, - Example 3-9 Obtain a fo;rnuln b r the electric field iltensity along the axis of a uniform line charge of length 4. The uniform Iioe-dmrgfi'densit is p,. --L

- Solution: For an infinitely lopg line charge, the E fieldtpn be determined readily by applying Gauss's law. Bs iq the solution to J3xamplp 3-4. However, for a line charge of finite length, as showq in Fig. 3-16, we cannot cqnstruct a Gaussian surface over which E . ds is constant. qpuss's law is therefore not pseful here.

Instead, we use Eq. (3-58)' by taking an element dlt = drf at il. The distance R from the charge elebent .to the p o i ~ t P(0, the axis of the line charge is . . , i . r

. , L . 6" R:= (Z z'), Z > T . .,

I

Here it is extremely important distinguish the podition bf the field point (unprimed coordinates) from the position ~f tpe source point (prime4 coordinates). We integrate

4 $.t'

3-6 (

ELECT

:3-61a)

3-61b)

1s very squarr

c1s of li

/'- r e t

a 11.- \L!Y~..

J'. The. hc l111e

~ n m e d regrate

n

3-6 / CONDUCTORS IN STATIC ELECTRIC FIELD 91 I? . '

, , . over the source region

. , pc ~ 1 2 dz' V = - 47E0 S-LIZ I - zf

z + (L/2) L ="In[: 1, 2,-. - 4nc0 z - (L/2) 2 ( 3 -62)

The E field at P is the negative gradient of V with respect to the unprimed field coordinates. For this problem,

The preceding two exampies iiiustrate the procedure for determining E by first finding V when Gauss's law cannot be conveniently applied. However, we emphasize that, i f~ymmetry conditions exist such that a Gaussian surfrrce cun be constructed over which E . ds is constant, it is always easier to determine E directly. The potential V , if desired, may be obtained from E by integration.

3-6 CONDUCTORS IN STATIC ELECTRIC FIELD

So far we have discussed only the electric field of stationary charge distrlbut~ons in frcc spacc or air. Wc now cxaminc: thc licld bchavior in matcnal mcdia. I n gcncral, we classify materials according to their electrical properties into three types: con- hc tors , setniconductors, and insulators (or dielectrics). In terms of the crude atomic model of an atom consisting of a positively charged nucleus with orbiting electrons, thc clcctrons in thc outermost shclls of Lhc atoms of~conductors arc vcry loosely held and migrate easily from one atom to another. Most metals belong to this group. The electrons in the atoms of insulators or dielectrics, however, are held firmly to their . orbits; they cannot be liberated in normal circumstances, even by the applicatlorl of an external elect~ic field. The electrical properties of semiconductors fall between those of conductors and insulators in that they possess a relatively small number of freely movable charges.

111 kcrms of khc band theory of solids, we find that there are allowed energy bands for electrons, each band consisting of many closely spaced, discrete energy states. Between these energy bands there may be forbidden regions or gaps where no eiec- trons of tbwlfd ' s atom can reside. Conductors have an upper energy band partially filled with electrons or an upper pair of overlapping bands that are partially filled so that the electrons in these bands can move from one to another with only a small change in energy. Insulators or dielectrics are materials with a completely filled upper band, so conduction could not normally.occur because of the existence of a large energy g?p to the next higher band. If the energy gap of the forbidden region is relatively small, small amounts of external energy may be sufficient to excite the electrons in the filled upper bapd to jump into the next band, causing conduction. Such materials are semiconductors.

I

92 STATIC ELECTRIC FlE+PS 3 ; > I ?

;.I The macroscopic electri 1 ptoperty of a mattrial r(lcdium is characterized by a

constitutive parameter calle~cdnhctitlity, which we Mil define in Chapter 5. The definition of conductivity, holueger$ is not importbt id this ;chapter because we are not dealing with current flow and are now interested anly in the behavior of static electric fields in material media,'In this sectiop we exgmi e the electric field and i charge distribution both insiqe t\iq bulk and on t4c surfi~cc qf a conductor.

Assumc for the prcscnl tkul yoiilc positive (oi qg+v@ichargc~ are introduced '

in the interior of a conductofi. An electric field will be let up in the conductor, the field exerting a force on the charges and making them nji,ve away from one another. This movement will continua'until all the charges reach the conductor surface and redistribute themselves ip suck a way that both thacharge and tfie field inside vanish. Hence, I ', I

(tJnCfer Static Copditions)

' 1 p = o -1 (3 -64)

1 ' I I

When there is no charge in the inteiior of a conquctgr (p ; 0ik must be zero because, according to Gauss's law, the lot@ putward electric flu* thropgh ony closed surface constructed inside the conduc or &st vanish. : t ,'

The charge distribution ob thesurface pf i, conductPr depends on the shape of the surface. Obviously the chqges ~ ~ o u l d not be in b state of eqdfjbrium if there were a tangential component of the electtic field intepsity that pro@ces k tangential force and moves the charges. Therefork; hnder sraric c o ~ d i t i o ' ~ the E3eld on a conductor surfice is everywhere normal tq th4sbrfaee. In othet:wor$, rh$ surface of a conducror is (In equrpofentiul surface u n $ r ydtic contlition.s.!A~ n,pqttct of fact, since E = O cvcrywlicrc inside o conclucior, f l ~ c whole col~d\rclor llasYthc w n c clcctro\tat~c potential. A finite limc is rc irql for thc chqrgcjs to f;pdistributc on a conductor surface and reach the eguilibri m State. This time qepenqp on the conductivity of the material. For a good c~pduct r s ~ c b as copper, this tim is in the order of 10- l 9 (s), J ? a very brief transient. (This ppinh will be elaborated in Section 5-4.)

Figure 3-17 shows an int fa* between a c~ncl$ctor $nd bee space. Consider the contour ubcdu, which hasew7ih'# = cd .= Aw and = Ah. Sides ab and cd are parallel to ihe int&fq$d Applying that E in a conductor is'tero, ' e bpiain immecjjavly

'

- ?' Y : & - : ; E ; .<' r' * $114 dl' = Et Av:= 0 : + J .,

or . . - L. (? E , = O , , , ! i !:a (3 -66)

which says that the tengential ~ot$nent of the $ fikld oLIh a ocnddq~toor surfoce is zero. In order to find E,,, the norm 1 cfiflponent of E ailthe fprfacc of the conductor, we

! ? , I > , !,' ;i --'

: l i . b' . L ' ( 4

I 1; i . t

; 1 3 !! ! ;*

:d by a 5. The we are

c f static :Id ayd

oduced or, the nother. ~ce and vanish.

,

. .

(3 -64)

( 3 -p, \

:cause, surface

! a l p of' :e were 11 force lducror ~;luc.tc,r E = O

O b l d t l C

lductor of the

- 1 9 (4,

der the des ub noting n

I

(3 -66)

is zero. tor, we

3-6 1 CONDUCTORS IN STATIC ELECTRIC FIELD 93 ,

1 r.* 1 6 , ) . A I

h 1 I . ; . . . I I I

I

i Ps I

1 . 'I

I

Fig. 3- 17 A conduct&-free space interface. I

construct a Gaussian surface in the form of a thin pillbox with the top face in free space and the bottom face in the conductor where E = 0. Using Eq. (3-7), we obtain

P E,, = -2. €0

Hence, the normal component of the E field at a conductor-free space boirilbry is equul to the surfuce churoe density on the conductor divided by the permirtioitp of j x e spucc. Summarizing the buundury condifions at the conductor surhcc, we have

When an uncharged conductor is placed in a static electric field. the external field will cause loosely held electrons inside the conductor to move in a direction opposite to that of the field and cause net positive charges to move in the direction of the f i e l q h e s e induced free charges bill distribute on the conductor surface and create an induc2d field in such a way that they cancel the external field both inside the conductor and tangent to its surface. When the surface charge distribution reaches an equilibrium, all four relations. Eqs. (3-64) through (3-67). will hold: ,

a'nd the conductor is agdin an equipotential body.

Example 3-10 A positive point charge Q is at the center of a spherical conducting shell of an inner radius Ri and an outer radius R.. Determine E and V as functions of the radial distance R.

I

94 STATIC ELECTRIC FIELDS ( 3 : , .

1:

t: ,' 1: 1; 1: *, L i

Fig. 3-14 J E I W ~ ~ field intensity and potential vdiatio$ of a point charge +Q at the knier flf a conducting shell ( ~ x a m ~ l e 3 , 0).

i , I 1 !

. li 1 :

Solution: The geometry qf the is shown I id' ~ g . j3- 18(a). Since there is spherical symmetry. it is sirnplest,jo u s i Gauss's law io determine E and then find V by integration. There are three p t i n a regions: (I) R > a, (b) R, S R 5 R,, and (c) R < R,. Suitable spherical Gaupiqn qurfaces will be cons{ructed in these regions. Obviously, E = a,E, in all fhree rbgiohd. - ! I

6 r

' - j 3 a) R > R, (Gaussian surface S,): i! i 1'

* i i! I! i

$ $ * i = E R , 4 n R 2 = - : jEO: B . or t

1- 3-

I

3-7' RL. E L E C ~ R I C

Idc e x t ~ ma1 eve effe

I 1 I

1

i

/'-

J

here is sn lind to, and .egions.

r.

(3 -68)

3-7 1 DIELECTRICS IN STATIC ELECTRIC FIELD 95 '

. The E field is ths kame as that o f a point charge Q without the presence of the shell. The potential referring to the'point at infinity is , -.. I

1 b) R , 5 R _< Ro (Gaussian surface S,): Because of Eq. (3-65), we L o w

ER2 = 0. (3 -70)

Since p = 0 in the conducting shell and since the total charge enclosed in surface S2 must be zero, an amount of negative charge equal to - Q must be induced on the inner shell'surface at R = R i . (This also means an amount of positive. charge, eqcl?! LO t Q is induced on the outer shell surface at R = R,.) The cull-

ducting shell is an equipotential body. Hence,

c) R < Ri (Gaussian surface S, ) : Application of Gauss's law yields the same formula for ER3 as ERl in Eq. (3-68) for the first region:

Q ER3 = - 4 x e , ~ ~ ' '

(3 -72)

The potential in this region is

& = -s ER3 dR + C = --- + c , 4m, R

where the integration constant C is determined by requiring V, at R = R, to equal V2 in Eq. (3-71). We have

and

The variations of E i and V versus R in all three regions are plotted in Figs. 3-18(b) and 3-18(c).

3-7 D IELECTRIC~N STATIC ELECTRIC FIELD

Ideal dielectrics do not contain free charges. When a dielectric body is placed in an external electric field, there are no induced free charges that move to the surface and make the interior charge density and electric field vanish, as with conductors. How- ever, since dielectrics contain bound charges, we cannot conclude that they have no effect on the electric field in which they are placed.

i F (6. !' , , * I .

96 STATIC ELECTRIC FIELDS / 3 ' , I I

dielectric material.

more dissimilar which do not have permanent cules arc of lhe order o f dipoles in a polar

on thelindividual that shown in

Fig. 3-19.

Int PO]

anc is

Her

Rec.

whe: of t t

1uclcus lectrlcs 3 f c p

30511, ,

1 S1T

! crc;llc; dlpoles !la1 the ~ d e the

t ' l C I I Ill

two or decules, r mole- hv~dual .lament livldual l o w in

n I l'LJcll-

(3 -74)

3-7 1 DIELECTRICS IN STATIC ELECTRIC FIELD 97 ,

: where n is the number of atoms per unit volume and the numerator represents the vector sum of the induced dipole moments contained in a very small volume Av.

. The vector P, a smoothed point function, is the volume density of electric dipole -S moment. The dipole moment dp of an elemental volume dv' is dp = P du', which

produces an electrostatic potential (see Eq. 3-48)

P a, d V = - dv' .

4 m 0 R Z (3-75)

Integrating over the volume V' of the dielectric, we obtain the potential due to the polarized dielectric. a

dv' , (3-76)+- L -

where R is the distance from the elemental volume do' to a fixed field polnt. In Cartesian coordinates,

RZ = ( X - x')' + ( y - y')2 + ( Z - zl)', ( 3 -77)

and it is readily verlfred that the gradient of 1/R with respect to the primed coordiacres is

(3-78)

Hence, Eq. (3-76) can be written as

Recalling the vector identity (Problem 2-18),

V 1 . ( f A ) = f v ' . A + A . v'f, . (3-80)

and letting A = P and f = l /R, we can rewrite Eq. (3-79) as

. 1 V' - P V = -[J, 4xc, V 1 - ( : ) d u r - sV, ~ d v ' ] . (3 -8 I )

The first volume integral on the right side of Eq. (3-81) can be converted into a closed surface integral by the divergence theorem. We have

where a. is the outward normal to the surface element ds' ofthe dielectric. Comparissn of the two integrals on the right side of Eq. (3-82) with Eqs. (3-57) and (3-56),

,

' We note here that V on the left sjde of Eq. (3-76) represents the electric potenriol at a field polnt. and V' on the right side is the volume of the polarized dielectric.

I . , 1

' * , . I i c $ $' .

1 . , . I

98 STATIC ELECTRIC FIELD^ 3 . . 1 1 ; .* " * . : ; ; J :.:-.,:. .;:...* . .

" - I I f , , . 1 : " :. ,

i # - respectively, reveals that the eleqplc pbtential (and thkrefoi the electric field intensity also) due to a polarized dielectriq can be calculated fiom the contributions of surface and volume charge distributions haying, respectivsld densities

,

and

(3-84)'

These are referred to as polariz(ltion,.charge densities or ((pund-charge densities. In other words, a polarized dieleetrio cnn,he rcplncrd hy oy rqsi~olent polorizntior~ a,$,ce dargc (lensity p, end U I I rquiuq~e~~t~polurizatioi~ voluflle charge density p p for field calculations. i

r

(3 -85)

with the aid of a vector identity, a charge distributions. . The sketch in Fig. 3-19 the ends of similarly oriented dipoles exist on fpolarization. Consider

an imaginary elemental T l~e application of an external electric field the bound charges: positive charge +q move a t h ~ field and negative charges - q move an equal thd field. The net total

4f2 F nq(d . anKAs). , , I '

(3 -86) ' tt But the dipole moment per 4 it volume, is by dehnitioR the polarization vector

P. We have . f ?, . ' t

9Q = P . an(As) ;+

9 ; J I (3-87) and t : ,

a,,, ! I

-. '

as given in Eq. (3-83). nbnnal. This relation

-w . . correctly gives a positive surface in Fig. 3-19 and a

negative surface charge on the I .

- - I" - 1 % ' The prune ~ g n oh a. md V h n heen drop@ br smplic~ty. rjnd'e' , 9s. 1;)-83) and (3-84) involve only

source coordinates and no confuqiqn ~)J'@u[t:! i' r 4 - ; , ,, : 3 -

i p ;: f 1 - r t i

1 . 1

> . $ 4 ' .J

i 5 i -

wit

wl1 whi to t the vcrl

whe

3-8 ELEC* DIELECTRIC

Bea the 1

diffe ' mils

tensity surface

(3-83)+

(3-84)'

11ca 111

surjace or field

(3 -rx

. ;I1

1111 ton$. llilldl.ly

on\~der II : \ I )

ll.ll;~cb:

2gallve et total .vhere 11

:ma1 to ld

(3-86)

I vector

(3-87)

p, '

r ~ I i \ t i \

9 and a

olve only

3-8 / ELECTRIC FLUX DENSITY AND DIELECTRIC CONSTANT 99

For a surface S bounding a volume V, the net total charge flowing out of V as a result of polarization is obtained by integrating Eq. (3-87). The net charge remaining

' within the volume V is the negative of this integral. L 0

Q = - $ s P . a,, ds I

- .

which leads to the expression for the volume charge density in Eq. (3-84) . Hence, when the divergence of P does not vanish, the bulk of the polarized dielectric appears ;; !x chdrged. However, since we started with an electrically neutral dielectric body, '

thc total charge of the body after polarization must remain zero. This can be readily verified by noting that

Total charge = $ p,,, (1s + Jv p, du

= $ s ~ . a , , d s - J v v . ~ d ~ = ~ , (3 -89)

where the divergence theorem has again been applied.

.8 ELECTRIC FLUX DENSITY AND DIELECTRIC CONSTANT

Ih.c:~\~sc :I ~xil;~ri/.ctl tliclcc~ric ~ ivcs rise t i ) : I vol~~nlc c h ; ~ r ~ c tlcnsily I),,. wc espcct ihc electric lield inicrlsity duc t o a given sourcc distribution in a ciiclcctric to be different from that in free space. In particular, the divergence postulated in Eq. (3 -4) must be modified to include the effect of p,; that is,

1 . -5' v.32 = d P + p,) . , (3-90)

€0

Using Eq. (3-841, we have

Wc now define a new fundamental field quantity, the electricJlux density, or e1ectr.i~ displacement, D, such that

The use of the vector D efiiiblcs us to write il divergence rc1:rtion hctwccn the electric field and the distribution of j i v e charges in any medium without the necessity of dealing explicitly with the polarization vector P or the polarization charge density p,. Combining Eqs. (3-91) and (3-92), we obtain the new equation

I !:i !' ; /

i ' . . { ,. f'

1 ,

I r. 1 . i where P is the voiume depsity C$ file; char& ~ ~ u a k i & s 13-;!I and (3 -5) are the two fundamental governing diffqreqfial equation9 ' f ~ f &ctrosta;ics in any medium. Note that thel~ermittivity af fred space, ro, does not! appaar e$licitly in these two equations.

$

The corresponding integral fyrrq ok Eq. (3-93) is obtaiqed by taking the volume integral of both sides. We have : '

; * ' i

I '

Equation (3-951, another f o p o f G&SS'S law, states! that ;he total outward pk ql the e/ecfric d i sp lacone~~ (or, simply, tlfe totol of~t~varif e 1 e e ; ~ i c ~ u s ) ouer ally closed suguce is equul to the total pee charg4 +enclosed in the~surjaqe. As-b,s been indicated in Section 3-4, Gauss's law is rqos{ju eful in dete&jninajthe elect& field due to charge distributions under symmetry c nditions. :' 6

When the dielectric p rope r fp b( the medium $re linear and isotropic, the polariztion is directly proporticjpa] to the electric geld iqensity, and the proportionality constant is independent of the direction of the field. We write

:i ,

, , # = EOXA (3 -96)

where x i is a dimensionless dielectric medium is linear if X. is of space

is a dimensionless constant know as t& relotiye or the dielectric constunt of the medium. The coefiicient c gsdil the absoluIe &@it jvity (often called simply

I permittiuit~) of the medium and 1s p,+sured in meter (F/m). Air has a dielectric constant of l.WQ59; habcei,tb ennittivity h us$lly taken as that of free space. The dielectric constants gf s f ! t t he r ma.r&i a r included in a table in Appendix B. : f

! 4 ' . L , . .. : I

; ! .!, ' 1 , 111 . ' A tensor would be required to represent the d&tric swceptibil ity~ the I i ' t - 'f i , :, . > :

is anisotropic.

So CO I

sol >f in , v 1

,the

- 5 t

the

:irc 111c ledium. '

cse two

volume

( 3 -94)

(3-95)

flu of close~f

t11c;llod d u i p

IIC. t' )rap .

(3 96)

I ~ ~ I U J I I

f spacc

1

(3-97)

(3 -98)

)rlstallf

sin* I L L

of fsw ~ b l e 1.

3-8 / ELECTRIC FLUX DENSITY AND DIkLECTRIC CONSTANT 101 . , . .

Nolo 1h;il r , can h a fur~clion ofspnm coordinnlcs. Ilr, is indcpcndcnt ofposition, the medium is said to be homogeneous. A linear, homogeneous, and isotropic medium is called~a simple medium. The relative permittivity of a simple medium is a constant.

I Example 3-11 A positive point charge Q is at the center of a spherical dielectric shell of an inner radius Ri and an outer radius R.. The dielectric constant of the shell is c,. Determine E, V, D, and P as functions of the radial distance R.

Solution: The geometry of this problem is the same as that of Example 3-10. The conducting shell has now been replaced by a dielectric shell, but the procedure of solution is similar. Because of the spherical symmetry, we apply Gauss's law to . 5nJ E and D inthiee regions: (a) R > R.; (b) Ri I R 5 and (cJ X <.A,. Potential V is found from the negative line integral of E, and polarization P is determined by the relation

P = D - cOE = E ~ ( E , - 1 ) E . ( 3 -99)

The E, D, and P vectors have only radial components. Refer to Fig. 3-20(a), whcre the Gaussian surfaces are not shown in order to avoid cluttering up the figure.

a) R > R, The situation in this region is exactly the same as that in Example 3-10. We have, from Eqs. (3 -68) and (3 -69),

Q v1 = -. 4nc,R

From Eqs. (3-97) and (3-99), we obtain

and P,, = 0.

The applicaiion of Gauss's law in this region gives us directly

Dielectric shell

Fig. 5-20 'Vicld yuri:itions of a point charge +& at the

. center of a dielectrjc shell (4 : : , (Example 3-1 1,) a

. . t 7 .. I

have a ,m.

3-8 I ELECTRIC FLUX DENSITY AND DIELECTRIC CONSTANT 1 J3

c) R < Ri Since the medium in this region is the same as that in the region R > R., the application of Gauss's law yields ;he 'same expressions for E,, DR,and PR in .... both regions:

To hnd v,, we must add to V2 at R = Ri the negative line integral of E,,:

The variations of e,E, and DR versus R are plotted in Fig. 3-?O(b). The difference ( D R - roE,) is PR and is shown in Fig. 3-PO(c). The plot for V in Fig. 3-20(d) is a composite graph for V,, V,, and V, in the three regions. We note that DR is a continuous curve exhibiting no sudden changes in going from one medium to another and that PR exists only in the dielectric region. It is instructive to compare Figs. 3-20(b) and 3-20(d) with, respectively, Figs. 3-18(b) and 3-18(c) of Example 3-! 1.

From Eqs. (3-83) and (3-84) we find

on the inner shell surface;

Q

--. on the outer shell surface; and

= - I ('

i . j aK.(K'l'R2) = 0. (3-lW)

Equations (3-107), (3-108);and (3-109) indicate that there is no net polarization volunle charge inside the dielectric shell. However, negative polarization surface charges exist on the inner surface; positive polarization surface charges, on the outer

!i - < 3 1 t 1

\ i 1 t I

1

104 STATIC ELECTRIC ~ 1 @ \ 0 ~ 1 ' 3 ' , ' : I I 3 ' .

I1

I r , 14 !

Table 3-1 qiclwf Strengths of ~ o ~ c ! ~ & n ~ o n Materials t . - + .

Material li ~ i c l k c t r i c ~ t r c n ~ t l ~ ( V / I ~ ) ----..------_I

) Air (atma~pher/c jyessure) . 3 x 10"

2

Mineral oil I ; l$ix lo6

Polystyrene . , i 2 0 x 1 0 ~ Rubber 25 )( lo6 Glass

1 L i 30 x 106

Mica , 200 x' 10" 7

I ,

surface. These surface charges produce an electric fisld intensity that is directed radially inward, thus reducing tbe.E field in region 2 due to the point charge + Q

3 > . . at the center. - - -_ I -- t

3-8.1 Dielectric Strength

We have explained that an g\ecf$c field causes small Pisplacements of the bound charges in a dielectric material, teiulting in pola&ation. If the electric field is very strong, it will pull electrons ~oppletely out of the m lecules, causing permanent dislocations in the molecular strtidure. Free char'ges w 11 appear. The material will 4 become conducting, and larle cbirents may res$lt. Tbis phenomenon is called a dielectric breakdown. Thi mqiqhA electric figld hten$Ity that a dielectric material can withstand without breakdqwn is the dielecttic st ength of the material. The approximate dielectric streng hs f some common8'$ubst~ce~are given in Table 3-1. f 9 The dielectric strength of a mqtendmust not be c&nfustid with its dielectric constant.

A convcnicnt number t o ~ c ~ k i n h c r i \ that tlsc iliu cctric strcnpth of ; i ~ r a t thc 1- atmospheric prcwirc 1s. 3 k Y / i ~ j ~ n . WIJCII L ~ I C S I C C ~ I I C I , I I C I C & I J I L C I I ~ A L Y cx~cccls Lll~s value, air breaks down.;,+4ass'ivivg~isnization taked'place, and sparking (corona dis- charge) follows. Charbe tendglto concentrate at sharp ~o in t s . In view of Eq. (3-67). the electric field intensity in t l ~ c immediate vicinitxd shgrp points is higher than that at points on a surface with i j small curvature. This is 'the principle upon which a

. lightning arrester works. Disgbay$e through the s h a ~ p oinL of a lightning arrester prevents damaging discharg~e thmugh nearby ~biec t - , 1 The fact that the electric field intensity tends to be point near the ~ s u of a charged conductor with a larger curvature is in the follo$ng

: I . I

-i * 7 Example 3-12 Consider twp sph:erical conductsjrs~ wjfh rqdii b, and b2 (b, > b,), which are connected by a co-ducdng wire. The ~$stange of,separation between the P conductors is assumed to be,yeqIarge cornpafed to ' so-that the charges on the spherical conductors may berco@iered as u a i f ~ ~ ~ l y ~ i s t r i b u t e d . A total charge Q

I J *

I I ( I

ii ; \ ! ! .

3-9 6 ELECTF

directed rg,: + Q

r---

: b~ , , , d ! is vfry i'nlancnt :rial will cailcd a Inatcrial -id. The :ble 3-1. :onstant. ir at the :cds this ona dis- . (3-67), han that which a arrester

: e l e p inductor

bz > bl), ween the :s on the :harge Q

3-9 1 BOUNDARY CONDITIONS FOR ELECTROSTATIC FIELDS

Fig. 3-21 Two connected conducting spheres (Example 3-12).

is deposited on the spheres. Find (a) the charges on the two spheres, and (b) the electric field intensities at the sphere surfaces.

Solution a) Refer to Fig. 3-21. Since the spherical conductors are at the same potential,

we have

Hence the charges on the spheres are directly proportional to their radii. But, since

we find

QI=- b1 Q and b2

b l + b2 2 - - - bl + b2 Q.

b) The electric field intensities at the surfaces of the two conducting spheres are

The electric field intensities are therefore inversely proportional to the radii, being higher at the surface of the smaller sphere which has a larger curvature.

3-9 BOUNDARY CONDITIONS FOR ELECTROSTATIC FIELDS

Electromagnetic problems often involve .hedia with different physical properties and require the knowledge of the relations of the field quantities at an interface between two media. For instance, we may wish to determine how the E and D vectors

/

w h

; ill!(

L , 0 , t

I Fig. 3-22 'An inttrface or between two medi~ . 1

l

ady know the bouqdary conditions that must interface. These copditi~ns have been given

I - whe

ln Eqs. (3-66) and (3-67). w e now consider an interface bl)tween&-general media shown in Fig. 3-22.

i

4 ' lnte Let us construct a smqll path abc% with sidesab and c m media 1 and 2 respec- the tlvely, both being parallel to t& inte'face and equal tb AW; ~ ~ u i t i o n (3-8), whlch is , becc assumed to be valid for regions containing discoqtihuouq~medja, is applied to this

path.' If we let s~des bc =ilo -Ah ,Tpproach zero, .their pontribufions to the line integral of E around the path can beiqglected. We hbve

i w h ~

"6,-' *f $ ebcde E . ~ ~ = ~ ~ . A w + ~ ~ . ( - A w ) = E ~ , A w ~ E ~ , I ; w = O .

I we 1 Therefore i

i I , t t ! or

(3-1 10) i t i I( cc

i I elec

which states that the tanyenrid cQppotaent of un E je4d is c~ntinuous acrosr an interface. Eq. (3-110) simplifies to &. (3-66) if one of the media is a conductor When media 1 and 2 are dieleftr&'$ith'.permittivities r; and r, respectively, we have

1 ,

' 1 4, ,, ;-=-. I . ) I

(3-1 1 1 ) i I

--\

if1 € 2 , .- i In order to find a relation betdecn the norrnal.com~pnems of the fields at a

- ,xa boundary, we construct a small illldx with its top ?ace ip medium 1 and bottom elec face in medium 2, as was ia Fig. 3-22. Thdfaces ave an area AS, and the

:i ,, . , fi height of the pillbox Ah is vanish nglj( small. ~ ~ ~ l ~ i n k Gayfs's law Eq. (3-95) to the Soil1

I $1 . L 1 . on: I I

See C. T Tal. -On the presentation of axwe$ theory: ~ r o c e e & ~ of (he IEEE, vol. 60. pp 936-945. "1 intc A U ~ U S ~ 1972. r , '1 necc

lat must .n given :I m@$

respec- \\ h1c i to this the line

t i n infur- r. When we have

(3-111)

/- :Ids a bottom

.and (j) to the

I ~ I 936-945,

3-9 / BOUNDARY CONDITIONS FOR FLECTROSTATIC FIELDS 137 4

i

pillbox, we have , I

$Dado = (Dl .an2 + D, .anl) AS i ' = a n 2 . (Dl - D,) AS

= pS As, (3-112)

where we have used the relation a,, = -a,,. Unit vectors a,, and a,, are, respectively, outward unit normals to media 1 and 2. From Eq. (3-112) we obtain

-1 (3-113a) 1

or . . . t

(3-113b)

a where the reference unit nornml is omvard from rnediurn 2. Eq. (3-1 13) states that the norntal component of D field is discontinuous across an

interface where a surface charge exists-the amount of discontinuity being equal ro -. the surface charge density. If medium 2 is a conductor, D, = 0 and Eq. (3-ll3b)

becomes Dln = 6lEln = P s ' (3-1 14)

which simplifies to Eq. (3-67) when medium i is free space. When two dielectrics are in contact with no free charges at the interface, p, = 0,

we have

Recapitulating, we find the boundary conditions that must be satisfied for static electric fields are as follows:

Tangential components, El, = E,,;

Normal components, a,, (Dl - D,) = ps.

Example 373- A lucite sheet ( E , = 3.2) is introduced perpendicularly in a uniform electric field E, = a,E, in free space. .Determine Ei, Di, and Pi inside the lucite.

Solution: We assume that the introduction of the lucite sheet does not disturb the original uniform electric field E,. The sitnation is depicted in Fig. 3-23. Since the interfaces are perpendicular to the electric field, only the normal field components need be considered. No free charges exist.

t * . \.d\ (

108 STATIC ELECTRIC F Lw;/ 3 lr 1, ' . r . . i

I i :I, 1 i

t

Boundary condition E ~ : (3-1;14) at the left ikt&rfagc gives

Di = a,D, = a,D, or

. $ 1 D, = axcoEo:

There is no change in electric )?@ density rcrpss thp in t&& The electric field intensity inside the ludte shbet iv

',I 1 , $ 4 F i = - D i = - D i = f l 3. ,e cOE, . +2

The polarization vector is zbro oqtside the lucitd sheet(Po = 0). Inside the sheet,

11 1 ~ , - ~ , i ~ ~ ~ ~ ~ = a . ( l $j)LO~O

=p a,b.k875r0~, (q/rn2)* I ! 2

Clearly, a similar applicati~n dffhe boundqry @ondipon Eq, (3-114) on the right interface will yield the prigifid 4 and D,, in ihdfree /paceon the right of the lucite sbcct. Dacs thc solution ofi(hi$ firohlcrn clli~n if tb dp ina l clcctric field i v n o t uniform, that is, if E, = q,E y) ' ) ,

, * * . $ , ,

. :t ?' ,

i 1 ; : . f <

i r . : . ? ;. . ( (; *

* . 1% : ,

I .

i i : 1 " [; i ' ;.

Fig. 3-24 ~dund+ conditions at the interfwe betw two dielectric media (E~a$~lc.?/!cl). ,

1 4

, I :. I I It

t 8 1 r

, I t 1 , t I

, i ; r 4 . . . 1 : i t !

hc ight J ' . 7~ luclte d is not

I 3-10 / C~PACITANCE AND CAPACITORS 109

I

,I ,!, Example 3-14. do dielectric media with Permittivities c, and c, are separated by a charge-fred boundary as shown in Fig. 3-24. The electric field intensity in mediuni 1 at the point F, has n magnitude E l and makks an angle c r , with the normal. Deter- mine the magnitude and direction of the electric field intensity at point P , in medium 2.

1

Solution: ~ w o , k ~ & t i o n s are needed to solve'for two unkpowns E2, and E,,,. After E2, and E,,, have been found, E, and I, will follow directly. Using Eqs. (3-1101 and (3 - 1 15), we have

. . . $ , . E, sin a, = El sin ai (3-117) and I n

h i ," ' c2E2 COS at = €,El cos b l . (3-118)

Division of Eq. (3-1 17) by 5q. (3-1 18) gives

-- tan a ,

The magnitude of E, is

By examinink Fig. 3-24, can you tell whether r , is larger or smaller than s,?

From Section 3-6 we understand that a conductor in a static electric field 1s an equipotential body and that charges deposited on a conductor will distribute themselves on its surface in such a way that the ilectric field inside vanishes. Suppose the potential due to a charge Q is V. Obviously, increasing the total charge by some factor k would merely increase the surface charge density p, everywhere by the same factor, witlrout-affecting the charge distribution because the conductor remains an equipotential body in a static situation. We may conclude from Eq. (3-57) that the potential of an isolated conductor'is directly proportional to the total charge on it. This may also bd seen from the fact that increasing V by a factor of k increases E = - V Y by a factor o f t . But, from Eq. (3-67), E = anp,/c,; it follows that p, and consequently the total charge Q will also increase by a factor of k. The ratio Q/V therefore

11 0 STATIC 'ELECTRIC FIELDS 1 3

remains unchanged. We write

where the constant of proportionality C is called the capacitance of the isolated conducting body. The capacitance is the electric charge that must be added to the body per unit increase in its electric potential. Its SI unit is coulomb per volt. or farad (F).

Of considerable importance in practice is the capacitor which consists of two conductors separated by free space or a dielectric medium. The conductors may be of arbitrary shapes as in Fig. 3-25. When a DC voltage source is connected'between the conductors, a charge transfer occuqs, resulting in a charge + Q on one conductor and - Q on the other. Several eiectric field lines originating from positive charges and terminating on negative charges are shown in Fig. 3-25. Notc tha! the field lines are perpendicular to the conductor S L I ~ ~ ~ I C C S . which arc cq~iipotcntiill sti~f;lces. Equation (3-121) applies here if V is taken to mean the potential ditrerence between the two conductors, V, , . That is,

The capacitance of a capacitor is a physical property of the two-conductor system. It depends on the geometry of the conductor6 and on the permittivity of the medium between them; it does not depend on either the charge Q or the potential difference V , , . A capacitor has a capacitance even when no voltage is applied to it and no free charges exist on its conductors. Capacitance C can be determined from Eq. (3-122) by either ( I ) assuming n V, , and dotermining Q in terms of V,,. or (2) assuming a Q and determining V , , in terms of Q. At this stage, since we have not yet

Fig. 3125 A two-conductor . capacitor.

i (3-121) 1

' . I r ,

isolated I

d to the I volt, or

! of two may be

3etween nductor ch wgcs he field urfaccs. )elween

r

( 3 - 1 ~

lductor y of the ~tcnrial :d to It d from . or (2) not yet

P

i ' 3

studied the methods lor solving boundary-value problems (which will be taken up in Chapter 4), we find C by the second method. he procedure is as follows:

1. Choose an apprdpriate coordinate system for the given geometry.

2. Assume chargks 4 Q and --Q on the conductors. 5

3. Find E from Q by Eq. (3-114)' Gauss's law, or other relations.

4. Find V I 2 by evaluating

from the condbctor carrying - Q to the other carrying + Q. 5. Find C by taking the ratio Q/V,,.

. - .- Example 3-15 A parallel-plate capacitor consists of two parallel conducting plates of area S separated by a uniform distance d. The space between the plates is filled with a dielectrlc of a constant permittivity E . Determine the capacitance.

Solution: A cross section of the capacitor is shown in Fig. 3-26. It is obvlous that the appropriate coordinate system to use is the Cartesian coordinate system. Follow- ing the procedure outlined above, we put charges + Q and - Q on the upper and lower conducting plates respectively. The charges are assumed to be uniformly distributed over the conducting plates with surface densities + p s and -p , , where

" From Eq. (3-1 14), we have

which is constant within the dielectric if the fringing of the electric field at the edges of the plates is negiectcd. Now

Fig. 3-26 Cross sectlon of a parallel-plate capacitor (Example 3-15).

Therefore, for a parallel-plate capacitor, .

I I - which is independent of Q or V, , .

For this problem we-could have started by assuming a potential difference V12 between the upper and lower plates. The electric field intensity between the plates is uniform and equals

The surface charge densities at the upper and lower conducting plates are + p , and -p , , respectively, where, in view of Eq. (3-67), ,

. 1'; 2 p., = e b , = c--.

d Therefore, Q = psS = (cS/(i)Vl2 and C = Q/V12 = sS/d. as before. - ---..- Example 3-16 A cylindrical capacitor consists of an inner conductor of radius a and an outer conductor whose inner radius is b. The space between the conductors is filled with a dielectric of permittivity c, and the length of the capacitor is L Deter- mine the capacitance of this capacitor.

Solutioti: W e use cylindrical coordinates for this problem. First we assume charges + Q and - Q on the surface of the inner conductor and the inner surface of the outer conductor, respectively. The E field in the dielectric can be obtained by applying Gauss's law to a cylindrical Gaussian surface within the dielectric a c r < b. (Note that Eq. (3-1 14) gives only the normal comporlent of the E field at a conductor surface. Since the conductor surfaces are not planes here, the E field is not constant in the dielectric and Eq. (3-1 14) cannot be used to find E in the u < r < b region.) Referring to Fig. 3-27, we have

Fig. 3-27. A cylindrical capacitor (Example 3- 16).

t p ; and

:hargcs c outer ~p i j ing ' . (Note ,urfacc. in the

ferring

3-10 1 CA~ACITANCE AND CAPACITORS 113

i

Again we neglect the fringing effect of the field hear the edges of the conductors. The potential difference Between the inner and outer conductors is

Therefore, for a cylin+ical capacitor,

We could not soive this problem from an assumed Vuh because the electric field is not uniform between the inner and outer conductors. Thus we wouid not know how lo cxprcss li and Q in lcrms of Vuh until wc l c m c d how to solve such a boundary- value problem.

Example 3-17 A sphcriwl capacitor consists of an inncr conducting spllcrc nl radius Ri and an outer conductor with a spherical inner wall of radius R,. The space in-between is filled with a dielectric of permittivity s. Determine the capacitance.

Solution: Assume ciiarges + Q and - Q, respectively, on the inner and outer conductors of the spherical capacitor in Fig. 3-28. Applying Gauss's law to a spherical Gaussian surface with radius R ( R i < R < R,,), we have

'Fig. 3-28 A spherical (Example 3- 17). '

capacitor

> .

i 11 4 STATlC ELECTRIC FIELDS 1 3

tiC?---+2y = fq Fig. 3-29 Series connection of

+ - 4- - capacitors.

Therefore, for a spherical capacitor,

3-10.1 Series and Parallel Connections of Capacitors

Capacitors are often combined in various ways in electrk-hcuits. The two basic ways are series and parallel connections. In the series, or head-to-tail. connection shown in Fig. 3-29.' the external terminals are from the first and last capacitors only. When a potential difTerence or electrostatic voltage V is applied, charge cumulauons on the conductors connected to the external terminals are + Q and - Q. Charges will be induced on the internally connected conductors such that +Q and -Q will appear on each capacitor independently of its capacitance. The potential differences across Ihc individual capacitors arc Q / C , , Q/Cz,. . . , (LIC,,. anti

where C, is the equivalent capacitance of the series-connected capacitors. We have

In the parallel connection of capacitors, the external terminals are connected to the conductors of all the capacitors as in Fig. 3-30. When a potential difference V is applied to the terminals, the charge cumulated on a capacitor depends on its capacitance. The total charge is the sum of all the charges.

-

' Capacitors, whatever their actual shape, are conventionaily represented in circuits by pairs of parallel bars.

on of

(3- 127)

/-

\ L O >IC

lnnef 'n mrs ",.,y. nuldt~ons rirgc\ w111 -Q will

~ITtrences

\\it Iiavc

( 3 - 128)

nected to krence V cis l r r s

of pdrallel

,I 3-10 / CAPACITANCE AND CA'PACITORS

I

F

I 1-Y- d -

b Fig. 3-30 Parallel connection + - of capacitors.

Therefore the equivalent capacitance of the parallel-connected capiicitors is

C , , = C , + C2 + . . . + C,,.

We note that the formula for the equivalent capacitance of series-connected capacitors is similar to that for the equivalent resistance of parallel-connected resistors and that the formula for the equivalent capacitance of parallel-connected capisitors is similar to that for the equivalent resistance of series-connected resistors. Can you explain this?

Example 3-18 Four capacitors C, = 1 pF, C2 = 2 pF, C, = 3 pF, and C, = 4 / I F are connected as in F ig 3-31. A DC voltage of 100 V is applied to the external terminals a-b. Determine the following: (a) the total equivalent capacitance between terminals u-b: (b) the charge on each capacitor; and (c) the potentla1 difirence across each capacitor.

Fig. 3-31 A cornblnatlon of 100 ( V ) capacitors (Exxmple 3- 18).

\

' 11 6 STATIC ELECTRIC FIELDS I 3

Solution

a) The equivalent capacitance C,, of C , and C , in series is

The combination of C 1 2 in parallel with C3 gives

C l Z 3 = C 1 2 + C 3 = 9 (PF). The total equivalent capacitance C,, is then

b) Since the capacitances are given. the voltages can be found as soon as the charges have been determined. We have four unknowns: Q,, Q,, Q,, and Q,. Four equations are needed for their determination.

Series connection of C , and C 2 : Q I = Qz.

Q i Qi'Q-3 Kirchhoff's voltage law, Vl + V2 = V3: - + - = -. c, cz c3

Q 3 Q4 Kirchhoff's voltage law, V3 + V4 = 100: - + - = 100. c, c4 Series connection at d : Q 2 + Q 3 = Q4.

Using the given values of C,, C,. C,. and C4 and solving the equations. we obtain

c) Dividing the charges by' the capacitances, we find

Q4 V4 = - = 47.8 (V). c4

These results can be checked by verifying That V, + V2 = V, and that V3 + V4 = 100 (V).

3-1 1 AND F

I.

-,

-

\ - d

~ h c charges Four equa-

r

,. \\ r: obtain

1 i 3-11 1 ELECTRdSTATlC ENEI'GY AND FORCES 117

4 I ,. I

I In Section 3-5 +. ihdicated that electrlc potential at a polnt in an eiectrlc field 1s 11

the work req$& rb bring a unit positive charge from mfinlty (at reference rero- L t potential) to that bolnt. In order to bring a charge Q, (slowly, so that kinetlc energy

and radiation effects may be neglected) from lnfinlty againrt the field of a charge Q, in free space to a distance R12, the amount of work required is

This work is stored in the assembly of the two charges as potential energy. Combining Eqs. (3- 130) and (3-131), we can write

Now suppose adother charge Q, is brought from.infinity to a point that is R , , from Q1 and R,, from Q,; an additional work is required that equals

The sum of AW in hq. (3-133) and W, in Eq. (3-130) is the potential energy. W,, stored in the asserhbiy of the three charges Q,, Q,, and Q,. That is.

We can rewrite W, ifi the following form:

--.- - + Q3 (" 4 7 r ~ ~ R , ~ ~ T ~ E ~ R ~ ~

= f (Qlv1 + QZV2 + Q3V3). (3- 135)

III Eq. (3-135), V,, tHe potential at the position of Q,, is caused by charges Qz and ,

Q,; it is different frbm the V, in Eq. (3-13n in the two-charge case. Similarly. V2 and V, are the potentials, respectiyely. at Q2 and Q, in the three-charge assembly.

Extending thii procedure of bringmg in ddditional charges, we arrive at the followins general expression for the potential energy of a group of N discrete point

I charges at rest. (The purpose of the subscript a on CC; is to denote that the energy is 1 1 of an electric nature.) We have I

I

where V,, the electric potential at Q,, is caused by all the other charges and has the following expression:

V,=- -. (3-137) 1 ( I * h l

j Two remarks are in order here. First, We can be negative. For instance, W2 in Eq. I

0-130) will be negative if Q, and Q, are of oppos~te signs. In that case, work is done by the field (not against the field) established by Q , in moving Q, from Infinity. Second, &, in Eq. (3- 136) represents only the lnteractlon energy (mutual energy) and does not ~nclude the work requlred to assemble the ~ndividuai polnt charges themselves (self-energy).

Example 3-19 Find the energy required to assemble a uniform sphere of charge of radius b and volume charge density p.

Solution: Because of symmetry. it is simplest to assume that the sphere of chake is assembled by bringing i ~ p a si~cccssion of sphsric;lI i:~yers of thicl\i>ess , / R , ~ c t l!~ili)l-ln v o l ~ l l l ~ C I I ; I ~ ~ C dcnrity he p, I\[ :I r:ldila H shorn in Fig. 3-32, the potential is

Q VR =-.!L-. 4 7 i ~ ~ R

where QR is the total charge contained in a sphere of radius R:

Q , ~ = pinl< 3 .

Fig. 3-32 Assembling a uniform sphere of charge (Example 3-19).

e energy is

(3- 136)

nd Ips the

(3- 137)

11; in Eq. )A 1s done m ~nfinity. ~lcrf- IIIJ rgcs u~cm-

f charge of

;f charge is K . Let the e potential

n

. , I '

The differential ch;rg& in a spherical layer of'thlckness dR is

and the work or energy in bringing up d Q ii

Hence the total work or energy rcquired to assbmble a uniform sphere of charge of radius b and chkrge dpnsity p is 'l

In terms of the total tharge 4 n ~

Q = p 7 b3, 3

we have

Equation (3-139) shows that the energy is directly proportional to the square of the total charge and inversely proportional to the radius. The sphere of charge in Fig. 3-32 could be a clodd of electrons, for instance.

For a continuchs charge distribution ofdensity p the formula for l.V, in L q (3- 136) for discrete charges must be modified. Without going through a separate proof, wz replace Q, by p dv and the summation by an integration and obtain

In Eq. (3-140), V is the potential at the point where the volume charge density is 17.

and V' is the volume of the region where p exists.

Example 3-20 Solvc the problcm in Example 3- 19 by using Eq. (3 - 140).

Soheion: In ~ x a m b l e 3- 19 we solved the problem of assembling a sphere of charge by bringing up a succession of spherical layers of a dilkrential thicknesb. Now we assume thatJte_ sphere of charge is already in place. Since p is a constant, it can be taken out of the integral sign. For a spherically symmetrical problem,

where V is the potehtial at a point R from the center. To find V at R, we must find the negative of the line integral of E in two regions: (1) E, = a R E R , from R = % to

120 STATIC ELECTRIC FIELDS I 3

R = b, and (2) E, = aRER, from R = b to R = 0. We have

and

Consequently, we obtain

Substituting Eq. (3- 142) in Eq. 13- 141). we get

which 1s the same as the result in Eq. (3- 138). Note that y, in Eq. (3- 140) includes the work (self-energy) required to assemble

the distr~bution of macroscopic charges, because it e the energy of interaction of every infinitesimal charge element w ~ t h all other infinitesimal charge elements. As a matter of fact. we have used Eq. 13- 140) in Emmpie 3 -20 to find the wif-energy of a un~form sphcr~c.~l charge. A, the radius h approaches zero, the self-energy of a (mathematical) point charge of a given Q is infinite (see Eq. 3-139). The self-energies of pomt charges Q, are not included in Eq. (3-136). Of course, there are, strictly, no I

point charges inasmuch as the smallest charge unit, the electron, is itselfa distrrbution I of charge.

3-11.1 Electrostatic Energy in Terms of Field Quantities

In Eq. (3-140), the expression of electrostatic energy of a charge distribution contains 1

the source charge density p and the potential function V. We frequently find it more I

, convenient to have an expression of We in terms of field quantities E and/or D. ,

without knowing p explicitly. To this end, we substitute V . D for p in Eq. (3-140):

W, = isv, ( V . D ) V d u .

Now. using the vector identity (from Problem P.2-18)

(3- 142)

r

I .isscmble :raction of lents. As a r-cncrgy of ncrgy uf a :If-energies strictly, no .istribution

t . jn contams ind i t inore m f ' ~ D,

:q. (3-140):

( 3 -43)

(3 - 144)

we can write Eq. (3-143) as

w . = + J V , v - ( v ~ ) d u - + J v , ~ . v v d u

- f $ s , V D . a , , d s + ~ , D - E d " , " s v (3 - 145)

-,

where the divergence theorem has been used to change the first volume integral Into a closed surface integral and E has been substituted for - V V In the second volume integral. Since V' can be any volume that includes all the charges, we miiy choose it to be avery large sphere with radius+k. As we let R -+ m, electrlc potential V and the magnitude of electric displacement D fall off at least as fast as, respect~vely, 1/R and l /RZ.+ The k e a of the boundmg surfate S' mcreaqes as R'. Hence the surface integral in Eq. (3-145) decreases at least '1s fast as 1/R and will v a u h ds K -+ x We are then left with only the second integral bn the right s ~ d a of Eq. (3-145)

1

/ W = ; J D - E d i ( J ) 1 ( 3 - 146'1)

Using the relation D = EE for a linear medium. Eq. (3-146a) can be wntten in two other forms:

and

We can always define an elecirostoiic energy density we mathematically, such that its volume integral equals the total electrostatic energy:

iq = Sv, we d c . (3-14;) We can, therefore, write

ul', = D . E (J/m3) (3 - 1 4Sn) or

---- ' For point charges V cc 1/R and D cr 1/R2; for dipoles' V cc 1/R2 and D cc l/R3

\

122 STATIC ELECTRIC FIELD? / 3

Fig. 3-33 A charged'parallel- - plate capacitor (Example 3-21).

However, this definition of energy density is artificial because a physical justification has not been found to localize energy with an electric field; all we know is that the volume integrals in Eqs. (3-146a, b, c) give the correct total electrostatic energy.

Example 3-21 In Fig. 3-33, a parallel-plate capacitor of area S and separation d is charged to a voltage V. The perhittivity of the dielectric is E . Find the stored electrostatic energy.

Soli~tion: With the DC source (batteries) connected as shown, the upper and lower plates are charged positive and negative, respectively. If thpf~inging of the field at the edges is neglected, the electric field in the dielectric is uniform (over the plate) and constant (across the dielectric), and has a magnitude

t l

Using Eq. (3-146b). wc havc

. ,

The quantity in the parentheses of the last expression, aS/d, is the capacitance of the parallel-plate capacitor (see Eq. 3-123). So,

Since Q = CV, Eq. (3-149a) can be put in two other forms:

and

I t so happensthat Eqs. (3-149a, b, c) hold true for any two-conductor capacitor (see Problem P.3-35).

I

I

stlfication .S that the energy.

raration d d electro-

md I ~ e r e 6;. dt he p' 7 )

*

ce of the

3-149a) t

1-149b)

0

1-149~)

pacitor

3-11 1 ELECTRil.'TATIC ENERGY AND FORCES 123

3-1 1.2 Electrostatic ~ o i c i s

Coulomb's law governs the force between two point charges. In a more complex system of charged qodies, using Coulomb's law to determine the force on one of the bodies that is caused by the charges on other bodies would be bery tedlous. This would be so even in [he simple case of fihding the force between the plates of a charged parallel~plkte capacitor. We will now discuss a method for calculating the force on an object iii a charged system from the electrostatic energy of the system. This method is based on the principle of virtudl displacement. We will consider two cases: (1) that of an,isolated system of bodies with fixed charges, and (2) that of a system of conducting bodies with fixed .

System of Bodies with Fixed Charges We consider an isolated system of charged conducting, as well Bs dielectric, bodies separated from one another with no connection to the outside world. The charges on the bodies are constant. Imagine that the electric forces Hkve displaced one of the bodies by a differential distance d t (a virtual displacement). The mechanical work done by the system would be

where F, is the total hectric force acting on the body under the condition of constant charges. Since we have an isolated system with no external supply of energy, this mechanical work mhst be done at the expense of the stored electrostatic energy: that is,

Noling from Eq. (2-LI I ) is Section 2--5 that the difkrential'change o ia scalar resulting from a position change dP is the dot product of the gradient of the scalar and l i t ,

we write d W, = (V We) * dl'. (3-152)

Since dP is arbitrary, con~parison of Eqs. (3-151) and (3-152) lends to

Equation (3-153) is & very simple formula for the calculation of FQ from the electrostatic energy of the system. In Cartesian coordinates, the component forces are

-1- 2 we ( F ~ ) x = -- 3s

(3 - 1 54a)

-, 124 STATIC ELECTRIC FIELDS 1 3

! If the body under consideration is constrained to rotate about an axis, say the

z-axis, the mechanical work done by the system for a virtual angular displacement ; d4 would be

dW = (T,): d 4 , i

(3-155) i

where (TQ), is the z-component of the torque acting on the body under the condition of constant charges. The foregoing procedure will lead to I

System of Conducting Bodies ivith Fixed Potentials Now consider a system where conducting bodies are held at fixed potentials through connections to such external sources as batteries. Uncharged dielectric bodies lnny also be present. A di'splacement de by a conducting body would result in a changc in total elcctrostatic energy and require the sources to transfer charges to the conductors in order to keep them at their fixed potentials. If a charge d Q , (which may be posiiive-or negative) is added to the kth conductor that is maintained at potential I/,, the work done or energy supplicd by the sources is V, dQ,. The total encrgy supplicd by the yoimxs to thc system is

The mechanical work done by the system as a consequence of the virtual displacement is

d W = F V . dt , (3-158)

where Fv is the electric force on the conducting body under the condition of constant '

potentials. The charge transfers also change the electrostatic energy of the system by an amount dWe, which, in view of Eq. (3-136), is

Conservation of energy demands that

Substitution of Eqs. (3-1571, (3-I%), and (3-159) in,Eq. (3-160) gives

F V . di? = dWe

= (V We) - dB or

uis, say the splacement

(3-155)

-. condition

(3-156)

1 ~ 1 1 1 hhcrc :h cxtcrnal placement :nergy and ,p them a t :, ,.%cd or emorgy cc the

(3-157)

' di$place-

(3-158)

~f constant he system

(3-159)

( -60)

(3-161)

3-11 / ELECTROSTATIC ENERGY AND FORCES 125

1

Comparison of E ~ S . (3-161) and (3-153) rekalr that the only difference between the formulas for the electric forces in the two cases is in the sign. It is clear that, if the conducting body 'is' constrained to rotate about the z-axis, the I-component of the electric torque will e b

which differs fronl Bq. (3-156) also only by a sign change.

Example 3-22 Determine the force on the conducting plates of a char, -ed parnilel- pliltc capacitor. Thc~plate\ have an ;ma S iind arc separated i n a i r by a d i ~ t r ~ i ; ~ \ . . Solutiot~: We solve the problem in two ways: (a) by assuming fixed charpcs: and then (b) by assummg fixed potentials. The fringing of field around the edges of the plates will be neglected.

Fixed ci1orge.s: With fixed charges 2 Q on the plates, an electric 6c!d intsnsity E, = Q/(coS) = V/.Y exists in the air between the plates regardless of their separation (unchanged by a virtual displacement). From Eq. (3-149b).

by,= i Q V = LQE 2 XI,

where Q and Ex are constants. Using Eq. (3-154a), we obtain

1 "(' ) 2 o2 : ( F y ) , = -- 5 QE,x = -- QE, = -A (3-163)

L X 2e0s1

where the negative signs indicate that the force is opposite to the direction of increasing x. It is an attractive force.

Fixed porenrials: With fixed potentials it is more convenient to use the expression in Eq. (3-149a) for bxY,. Capacitance C for the parallel-plate air capacitor is eOS/x. We have, from Eq. (3-161),

How different are (FQ) , in Eq. (3-163) and (F,), in Eq. (3-lM)? Recalling the relation

-\ ---- we find

The force is the same in both cases, in v i t e of the apparent sign difference in the formulas as expressed by Eqs. (3-153) and (3-161). A little reflection on the physical problem will convince us that this must be true. Since the charged capacitor has fixed dimensions, a given Q will r'esult in a fixed V, and vice versa. Therefore there 1s

I

i 126 STATIC ELECTRIC FIELDS I 3 !

t a unique force between the plates regardless of whether Q or V is given. and the 1 force certainly does not depend on virtual displacements. A change in the conceptual I constraint (fixed Q or fixed V ) cannot change the unique force between the plates. I

The preceding discussion holds true for a general charged two-conductor capacitor with capacitance C. The electrostatic force F, in the direction of a virtual displacement d t for fixed charges is i

Q2 dC

For fixed potentials,

V 2 iC QZ JC (3-167)

It is clear that the forces calculated from the two proccdurcs. which assumed different constraints imposed on the same cb;~r& c;lp;lcil~r. ;ire equal.

-.- \.

REVIEW QUESTIONS

R.3-I Write the dillcrential form of the fundamental postulatcs of clcctrostatics in free sp;lcc,

R J - 2 Under what conditions will the electric field intensity be both solenoidal m d irrotntional?

RJ-3 Write the integral form of the fundanlental postuiates of electrostatics in free space. and state their meaning in words.

R 3 - 4 When the formula for the electric field intensity ofa point charge, Eq. 13-12). w;ls derived,

a) why was it necessary to stip~llate that q is in a boundless free space? b) why did we not construct a cubic or a cylindric~~l surface ;,round y!

R.3-5 In what ways does the electric field intensity vary with distance for

a) a point charge? b) an electric dipole?

R.3-6 State Coulomb's law. . R.3-7 State Gauss's law. Under what conditions is Gauss's law especially useful in determining the electric field intensity of a charge distribution?

R.3-8 Describe the ways in which the electric Reid intensity of an infinitely long, straight line . charge of uniform density varies with distance?

R.3-9 Is Gauss's law useful in finding the E field o l a finite linichrrges! Explain.

R.3-10 See Example 3-5, Fig. 3-8. Could a cylindrical pillbox with circular top and bottom faces be chosen as a Gaussian surface? Explain.

R 3 - 1 1 Make a two-dimensional sketch of the electric field lines and the equipotential lines of a point charge.

1, and the :onceptual : plates.

or capaci- 1 displace-

(3-166),

lO7)

I dill'crent

f - ,

ii'cc \.p3ce.

o:~tional?

spacc, and

IS deri\ ed.

.sight line I r'

lines of a

I REVIEW QUESTIONS 127 I . i

R.3-12 At what valuk of 6 is the E field of a z-dirdcted electric dipole pointed in the negative z-direction?

R3-13 Refer to Eq. (3-59). Explain why the absoldte sign around 2 is required.

R.3-14 If the electric potential a t a point is zero, does it follow that the electrical field intensity is also zero at that poiht? Explain.

R3-13 If the electric held intensity at a point is zero, does it follow that the electric potentiai is also zero at that point? Explain.

R.3-16 An uncharged spherical conducting shell af a finite thickness is placed in an external electric field E,, what is the electric field intensity at the center of the shell? Describe the charge distributions on both the outer and the inncr s u r f x u of the shell.

.. !<.' . 17. ( ' : i l l V'( l/lt) 111 Iiq. (3 79) lx i~c~)laccd by V( i,/1(jt! LxpIai~i.

R.3-18 Define polariztltion vector. What is ~ t s SI unit?

R.3-19 What are polarizat~on charge dcnsities? What are the SI units Tor P . a,, and V . P ?

R.3-20 What do we rflean by simple inerlium?

R.3-21 Define electrrc displucei~ei~t wcior. What is its SI unit?

R.3-22 Define electric susceprihility. What is its unit?

K.3-23 What is the differcncc bctwecn thc permittivity and the dielectric coilmnt of a mechum'!

R3-24 Does the electric flux density due to a given Charge distribution depend on the properties of the medium? Does the electric field intensity?

R.3-25 What is the difference between the riielectric constunt and the dielectric strengrh of a dielectric material?

R.3-26 What are the kenera1 boundary conditions for electrostatic fields at an interface between two different dielectric media?

R.3-27 What are the boundary conditions for electrostatic fields at an interface between a conductor and a dielectric with permittivity c ?

R.3-28 What is the boundary condition for electrostatic potential at an interface between two different dielectric media'!

113-29 Does a force exist between a point charge and a dielectric body? Explain.

R.3-30 Define cupacitance and cupucitor. - -. R.3-31 Assume that the permittivity of the dielectric in a parallel-plate capacitor is not constant. Will Eq. (3-123) hold if the average value of permittivity is used for E in the formula? Explain.

R.3-32 Given three lbpF capacitors, explain how they should be connected in order to obtain a total capacitance of

128 STATIC ELECTRlC FIELDS I 3

R3-33 What is the expression for the electrostatic energy of an assembly of four discrete point ' charges?

R3-34 What is the expression for the electrostatic energy of a continuous distribution of charge in a volume? on a surface? along a line?

R.3-35 Provide a mathematical expression for electrostatic energy in terms of E and/or D.

R.3-36 Discuss the m ~ a n i n g and use of the principle of virtual displacement.

R3-37 What is the relation between the force and the stored energy in a system of stationary charged objects under the condition of constant charges? under the condition of fixed potentials?

.' PROBLEMS ,

P.3-1 Refer to Fig. 3-3.

a) Find the relation between the angle of arrival, a, of the electron beam at the screen and the deflecting electric field intensity Ed.

b) Find the relation between wand L such that d , = d0,'2O. -,.- 1'3-2 Tllc ca[liotlc-v:ly oscillogl.;~ph (CNO) sliown in Fig. 3 3 is used to nicasurc tlic \:OII;IFL' applied to the paralid deflection plates.

a) Assuming no breakdown in insulation, what is the maximum voltage that can be measured if the distance of separation between the plates is l ~ ?

b) What is the restriction on L if the diameter of the screen is D? c) What can be done with a fixed geometry to double the CRO's maximum measurable

voltage?

P3-3 Calculate the electric force between the electron and nucleus of a hydrogen atom. assuming they are separated by a distance 5.28 x lo-'' (m).

P3-4 Two point charges, Q, and Q,, are located at (1,2,0) and (2,0, O), respectively. Find the relation between Q, and Q,, such that the total force on a test charge at the point P(- 1, 1 , O ) will have

a) no x-component, b) no y-component.

P3-5 Two very small conductiig spheres, each of a mass 1.0 x (kg) are suspended at a common point by very 'thin nonconducting threads of a length 0.2 (m). A charge Q is placed on each sphere. The electric force of repulsion separates the spheres, and an equilibrium is reached when the suspending thread makes an angle of 10'. Assuming a gravitational force of 9.80 (Nfig) and a negligible mass for th'e threads, find Q.

P3-6 A line charge of uniform dcnsity p, in free space forms a semicircle of radius b. Determine the magnitude and direction of the electric field intensity at the center of the semicircle.

P.3-7 Three uniform line charges-p,,, p,,, and p,,, each of length L-form an equilateral triangle. Assuming p,, = 2p,, = 2pt3, determine the electric field intensity at the center of the

a triangle. h

hscrete point

on of charge

and/or D

)f 5tstiunary i potentials?

: 5~:ucn and

~ h e p p g e

1 1 L 1-

measurable

1 atom, as-

). Find the 1, 1, 0) will

)pended at pi.iced on !r redchcd

1 ' . * t - i r

n D r t ~ le

quliaL ..I ~ter of the

PROBLEMS 129

P.3-8 .4ssuming that h e electric field density is E 4 a,100x (V/m), find the t a l e c t r i c charge contamed insidc ' 1

a) a cubicai v h n c 100 (mm) on a side centerdd at the ongin. b) a cylindricii vdlume bf radius 50 (mm) and height 100 (mm) centered at the origin.

? 1. P.3-9 A spherlcai didtributlon of charge p = p , / [ l - (R2ib')] exlsts in the regron 0 5 R 5 b Thls charge distrlbutldn is concentrically surrounddd by a conductmg shell wlth Inner radlus R, (> b) and outer radibs R,. Determine E everywhet'e.

P.3-10 TWO infinilely.long coax~al cylindrical surfdces, I- = a and r = b ( b > a), carry surface charge densities p,, ~ n d p,, respectively.

I

bctwccn o and b in order that E van~>hc\ for r > b'! -

P.3-I I At what va!uus ot' 0 docs the electric field inten,lty of a xiirccted d~pol r have no z- component?

a) Determine V d;d E at a distant point P ( R , 0, (b). b) Find the equations for equiporential surfaces and streamlines C) Sketch a famil? of equipotential lines and streamlines.

(Such an arrangement of three charges is called a linear electrostatic qimdrupolr.) I*

P.3-13 A finite line chsrge of length L carries a uniform line charge density p , ,

a) Determine V iil thr plane bisecting the line charge. b) Determine E from p , directly by applying Coulomb's law. c) Check the answer in part (b) with -VV.

P.3-14 A charge Q is distributed uniformly over an L x L square plate. Determine I/ and E at a point on the axis perbendicular to the plate, and through its center.

P.3-15 A charge Q is distributed uniformly over the wall of a circular tube of radius b and height h . Determine V and E on its axis

4) at a point outside the tubs, then b) at a point inside the tube.

P.3-16 A simple clasdcsl model of an aton1 consists of a nucleus of a posit~ve charge k ~ r ~ surrounded by a spherical electron cloud of the same total negative charge. ( N is the atomic number and e is the electronic charge.) An external electric field E,, will cause the nucleus to be displaced a distance r,, from the center of the electron cloud, thus polarizing the atom. Assuming a uniform ch~~e-d is t r ibu t ion within the electron cloud of radius b, find r,,.

P.3-17 Determine the work done in carrying a -Z(pC) charge from P,(2, 1. - 1) to P,(8. 2, - 1) in the field E = a,y + a,,x

a) along the parabola s = ~ J J ~ ,

b) along the straight line joining P, and P2

130 STATIC ELECTRIC FIELDS I 3

P.3-18 The polarization in a dielectric cube of side L centered at the origin is given by P = P,(a,x + a,y + a,z).

a) Determine the surface and volume bound-charge densities. b) Show that the total bound charge is zero.

P3-19 Determine the electric field intensity at the center of a small spherical cavity cut out of a large block of dielectric in which a polarization P exists.

P.3-20 Solve the folloiving problems:

a) Find the breakdown voltage of a parallel-plate capacitor, assuming that conducting plates are 50 (mm) apart and the medium between them is air.

b) Find the breakdown voltage if the entire space between the conducting plates is filled with plexiglass, which has a dielectric constant 3 and a dielectric strength 20 (kV,'mm).

c) If a 10-(mm) thick plexiglass is inserted between the plate.;, what is the maximum voltage that can be applied to the platCs without a breakdown?

P3-21 Assume that the z = 0 plane separates two lossless dielectric regions with E , , = 2 and E , ~ = 3. If we know that E, in region 1 is a,2y - a,.3.u + a,(5 + z), what do we also know about E, and D, in region 2? Can we determine E, and D, at any point ip region 2? Explaln. -._ _ -

, P.3-22 Dctcrtninc t i~c bound:~ry condirions for thc tangential and the normal components of P at an interface between two perfcct diclcctric tncdia with dielcctric constants E , , 2nd E , ? .

P.3-23 What are the boundary conditions that must be satisfied by the electric potential at an interface between two perfect dielectrics with dielectric constants E,, and E,, ?

P.3-24 Dielectric lenses can be used to collimate electromagnetic fields. In Fig. 3-34, the left surface of the lens is that of a circular cylinder, and the right surface is a plane. If E, at point P(r,, 45", 5 ) in region 1 is a,5 - a,3, what must be the dielectric constant of the lens in order that E, in region 3 is parallel to the x-axis?

Fig. 3-34 Dielectric lens (Problem P.3 -24).

P3-25 The space between a parallel-plate capacitor of area S is filled with a dielectric whose permittivity varies linearly frome, at one plate (y = 0) to E , at the other plate (y = d). Neglecting fringing effect, find the capacitance.

,by P =

out of a

nducting

, l i t ( \ with 11:11j.

n vduge

= 2 and .w about

r-

.>ncnt,- - f

1d Er*

%ti31 ar an

4. the left , st point ; in order

n

xric whose Keglecting

I PROBLEMS 131 ,

P.3-26 Consider the earth as a conducting sphere of radius 6.37 (Mm).

a) Determine its ~ a ~ a c i t a n c e . b) Detcrminc thimhximum charge that can exist on it without cawing a breakdown of the

air surroundirlg it.

P.3-27 Determine the capacitance of m isolated cbnducting sphere of radius h that is coated with a dielectric layer ~f uniform thickness d The dielectric has an electric susceptibility %. .

P3-28 A capacitor cbnrists of two concentric spnerical shells of radii R, add Ro. The space between them is filled with a dielectric of relative pkrmittivity c from R, to b(R, < b < R.) ard another dielectric of relative permittivity 26, from b to R,.

a) Determine E and D everywhere in terms of An applied voltage I/. h) Dctcrminc thc c:~paci tancc.

P.3-29 Assume that the outer conductor of the cylindrical capacitor in Example 3-16 is grounded, and the inner conductor is maintained at a potential Vn.

a) Find the electiic field intensity, E(a), at thcsurface of the inner conducror. b) With the irrner radius, b, of the outer conductor fixed, find a so that E ( n ) is minimized. c) Find this minimum E(ul. d) Detcrrhine the capacitance under the conditions of part (b).

P.3-30 The radius of the core and the inner radius of the outer conductor of a very long coaxial transmission line are ri and r, respectively. The space between the conductors is filled with two coaxial layers of dielectrics. The dielectric constant4 of the dielectrics are E,, for ri < r < b and E , ~ for b < r < ro. Determine its capacitance per uriit length.

P.3-31 A cylindrical capacitor of length L consists of coaxial conducting surfaces of radii ii

and r,,. Two diebctricmcdii~ of dillcrcnl diclvctric copstants r , , and r,, fill the space between the conducting surfaccs as shown in Fig. 3-35. Detcrmihe its capacitance.

- A Fig. 3-3 I - with two dielectrl

(Problem P.3-31).

5 A cylindrical capacitor ic media

P3-32 A capacitor consists of two coaxial me@llic cylindrical surfaces of a length 30 (mm) and radii 5 (mm) and 7 (mm). The dielectric material between the surfaces has a relative permittivity E, = 2 + (4/r), where r is measured in mm. Determine the capacitance of the capacitor.

132 STATIC'ELECTRIC FIELDS I 3

t P.3-33 Calculate the amount of electrostatic energy of a uniform sphere of c h a m with radius b and volume charge density p stored in the following regions:

a) inside the sphere, b) outside thesphere.

Check your rekits with those in Example 3-19.

[ P.3-34 Find the electrostatic energy stored in the region of space R z 6 around an electric dipole oT moment p.

I

P.3-35 Prove that Eqs. (3-149) for stored electrostatic energy hold true for any two-conductor ,: capacitor. i

P.3-36 A parallel-plate capacitor of width s, length L. and separation d is partially filled with a dielectric medium of dielectric constant t.?m shown in Fig: 3-36. A battery of I,, volts is connected between the piates.

a) Find D. E, and p, in each region. b) Find distance r such that the electrostatic cnergy storcd in each region i r the same.

Fig. 3-36 A parallel-plate capacitor (Problem P.3-36).

P.3-37 Using the principle of virtual displacement, derivc m expression for the force between two point charges +Q and -Q separated by a distance x in free space.

1

P.3-38 A parallel-plate capacitor of width IV. length L. and separation d has a solid d~electric '

slab of permittivity r in the space between the plates. The capacitor is charged to a voltage Yo by a battery, as indicated in Fig. 3-37. Assuming that the dielectric slab is withdrawn to the position . shown, determine the force acting on the slab

a) with the switch closed, b) after the switch is first opened.

\ Switch

. . .L .. . x Fig. 3-37 A partially filled parallel-plate capacitor (Problem P.3-38).

1 i " I

< ,

with rad~us I

an electric

)-conductor

filled w ~ t h a 011s IS con-

.ce between

d diclectric ~Itagc V, by hi position

4-1 INTRODUCTION

Electrostatic problems are those which deal with the effects of electric charges at rcst. These problem,d can present themselves in several different ways according to what is initially kiwwn. Thc solutioli usually calls for ~ h c determination of electric potential, electric field intensity, and/or electric charge distribution. If the charge distribution is given, both the electric potentilll and the electric field inlensity can

- be found by the formulas developed in Cha ter 3. In many practical problems, however, the exact qharge distribution is not everywhere, and the formulas in Chapter 3 cannot be applied directly for finding.the potential and field intensity. For instance, if the charges at certaln discrete points in spaco and the potentials of some conducting boclics arc givcn, it is rathcr difficult LO find the distribution of surface charges on the conducting bodies and/or the electric field intensity in space. When the cobducting bodies have bohdaries of a simple geometry, the method of images may be used to great advantage. This method will be discussed in Section 4-4.

In another type: of problem, the potentials of all conducting bodies may be known, and we wish to find the potential and field intensity in the surrounding space as well as the distribution of surface charges on the conducting boundaries. Differential equations must be solved subject to the appropriate boundary conditions. The techniques far solving partial aiifer-ctjai equations in the various coordinate systems.wil1 be discussed in Sect;& 4-5 through 4-7.

* .

. ,

4-2 POISSON'S AND LAPLACE'S E Q U A T I O ~ ~

In Section 3-8, we pointed out that Eqs. (3-93) and (3-5) are the two fundamental gpverning dificrcntirll cquations for clectrostatics in any medium. These equations are repeated below for convenience. ,

Eq. (3-93):

Eq. (3-5):

134 SOLUTION OF ELECTROSTATIC PROBLEMS 1 4 "' ' ' '"?

The irrotational nature of E indicated by Eq. (4-2) enables us to define a scalar electric potential V, as in Eq. (3-38).

In a linear and isotropic medium, D = rE, and Eq. (4-1) becomes

V . c E = p . (4-4)

Substitution of Eq. (4-3) in Eq. (4-4) yields

where E can be a function of position. For a simple medium$ that is, for a medium that is also homogeneous, E is a constant and can then be taken out of the divergence operation. We have ,

- - - . . In Eq. (4-G), we have introduced a new operator, V2, the Luplnciun operutor, which stands for "the divergence of the gradient of," or V V. Equation (4-6) is known as Poisson's equution; it states that the Laplacian (the divergence of the gradient) of V equals - p/e ji,r u simple tnediutn, where e is the permittivity of the medium (which is a constant) and p is the volume charge density (which may be a function of space coordinates).

. Since both divergence and gradient operations involve first-order spatial derivatives, Poisson's equation is a second-order partial difkrential equation that holds at every point in space where the second-order derivatives exist. In Cartesian coordi- nstcs,

and Eq. (4-6) becomes

Similarly, by using Eqs. (2-86) and (2-102). we can easily verify the following expressions for V2V in cylindrical and spherical coordinates.

Cylindrical coordinates:

(4-3) *

(4-3)

(4-5)

1 mcdiLm livergence

(4-6) /-

tor. \, ..dl known as -adient) of urn (which t l of space

:la1 Jeriva- at holds at an coordi-

!'I ; - /

(4-7) n

Ilowh k-

(4-8)

4 %

' ' ~ 4 - 2 ' 1 PO IS SON*^ AND LAPLACE'S EQUATIONS 135 ' r I

- < . .

1 d'V (4-9)

The solut~on of Poidon's equation in three dimensions subject to prescribed boundary conditions is,.ih beneral, not an easy task.

At points ifi a kimple medium where there is no free charge, p = 0 and Eq. (4-6) reduces to Y

which is known as Laplace's eqr~atiotr. Lapiace's equation occupies a very important position in electrom~netics. It is the governing equation for problems involving a set of conductors, sbch as capacitors, maintained at different potentials. Once V is found from Eq. (4-1 o), E can be determined from - VV, and the charge distribution on the conductor sdrfaces can be determined from p , = € E n (Eq. 3-67). .

I Example 4-1 The .two plates of a parallel-blate capacitor are separated by a distance d and mainiained at potentials 0 and V,, as shown in Fig. 4-1. Assuming negligible fZlnging effect at the edges, determine (a) the potential at any point between the plates, and (b) the surface charge densities at the plates.

a) Laplace's equation is the governing equatidn for the potential between the plates since p = 0 therq. Ignoring the fringing effect of the electric field is tantamount to assuming that tHi field distribution between the plates is the same as though the plates were infidtely large and that there is no variation of V in the x and z directions. Equation (4-7) then simplifies to

where d2/dy' is used instead of d-,;!'. since y is the only space variable.

-1- ' t

Fig. 4-1 A parallel-plate capacitor (Example 4-1).

t

136 SOLUTlON OF ELECT .ubr''!C PROBLEMS / 4 i

f Integration of Eq. (4-11) with respect to y gives , . ; dV - = C1, dy I

where the constant of integration C, is yet to be determined. Integrating again. we obtain

v = Cly + C,. (4-12) I

1 Two boundary conditions are required for the determination of the two constants of integration: I

Substitution of Eqs. (4-13a) and (4-13b) in Eq. (4-12) yields immediately C , =

V, /d and C, = 0. Hcnce thc potmllid at m y point I. bctwern 111~. plates is. from Eq. (4-12),

The potential increases linearly from y = 0 to y = d.

b) In order to find the surface charge densities, we must first find E at the conducting ,

plates at y = 0 and y = d. From Eqs. (4-3) and (4-14), we have

\

which is a constant and is independent ofy. Note that the direction of E is opposite to the direction of increasing V. The surface charge densities a t ihe conducting plates are obtained by using Eq. (3-67)'

1' I(, -> :I,, . 11; - * E

At the lower plate,

'-7

At the upper plate,

Electric field lines in an electrostatic field begin from positive charges and end in negative charges.

g again,

(4-12)

mstants

(4- 13a) (4-4 3b)

.ly C, = IS, from

( @ P I )

~duc t~ng

(4-15)

opposite nductmg

n \

; and end

t 1 Ekample 4-2 Detehine the E field both in~ide hnd outside a spherical cloud of electrons with a unibrm volume charge density p = - p , for 0 i R 5 b and p = 0 for R > b by solving~Poisson's and Laplace's equations for V.

I

Solution: We re$ that this problem was dblveh in Chapter 3 (Example 3-6) by applying Gauss's aw. We now use the sam problem to illustrate the solut~on of one-dimensional Poisson's and Laplace's equ tions. Since there are no variations 3 in 0 and 4 directio?s,'tue are only dealing with functions of R in spherical coordinates.

J a) Inside the cloud,;

-1.

t O S R I b , p y - P O .

In this region, Poisson's equation ( V 2 y - p h o ) holds. Dropping i l Z O and a/@ terms from kq. (4-9), we have

which reduces to

Integration of E4, (4-16) gives

The electric field intensity inside the electron cloud is

Since Ei cannot be infinite at R = 0, the integration constant C, in Eq. (4-17) must vanish. We obtain

P 0 E i = - a , - - R , 0 1 R s b . (4-18) 360

b) Outside the cloud, R 2 b , p - 0 .

Laplace's equaiion holds in this region. We have V2 Y. = 0 or

Integrating Eq. (4-191, we obtain

138 SOLUTION OF ELECTROSTATIC PROBLEMS 1 4 i

The integration constant C , can be found by equating Eo and E, at R = b, where t there is no discontinuity in medium characteristics. i

c2 P o -- -- b2 kO b , i

from which we find pOb3

C2 = -- (4-22) 1

3e0 4 -3 and ELECT

P

' O b 3 R R ~ . Eo = -a R- 3c,R2 '

(4-23)

Since the total chargc coniaincd in the clcctron cloud is

Equation (4-23) can be written as .

which is the familiar expression for the electric field intensity at a point R from a point charge Q.

Further insight to this problem can be gained by examining the potential as a function of R. Integrating Eq. (4-17), remembering that C , = 0, we have

It is important to Aote that C; is a new integration constant and is not the same as C,. Substituting Eq. (4-22) in Eq. (4-20) and integrating, we obtain

However, C; in Eq. (4-26) must vanish since V, is zero at infinity (R -+ a). As electrostatic potential is continuous at a boundary, we determine C; by equating l/i and Voat R = b:

p0b2 ; b'+C - pob2 660

1 - -- 3c0

(4-22)

(4 - 23)

n

(4-24)

t R from

itial as a

(4-25)

same as

(4-26) r'

elec~ : T/T anu

1 4-3 1 UNIQUENESS Q+ ELECTROSTATIC SOLUTIONS 139

4 i' , .

3 ..or r . ' 5 . .. .. * C

: p o d Z . - c; = --, 2%

(4-27)

'and, from Eq. (4-25),

We see that C: in &q. (4-28) is the same as V in Eq. (3-i42), with p = - p o . P I

11 4-3 UNIQUENESS OF irh

ELECTROSTATIC SOLUTIONS

111 L I I C Lwu rcla~ivcl~. siniplc cxa~nplcs ~n t l~e Ids[ section, wc ob1.1' ' lned t i x solutions by direct integration. In more complicated situations other methods of solution must be used. Befoie these methods are discussed. it is important to know that n solutioti o j Poissoft's eq~ut ion (of which Laplace's equation is s special cxci r h r rurisjies rile giaen hoiiniiaq. c~oiiilirioi~s is ii wiipzrr .solsiioti. This statement is cnlied the u~liqtieness theorem. The implication of the uniqueness theorem is that a solution of an electrostatic problem with its bou~ldsry conditions is tile o~ l i ! poisihle . s ~ / ~ i t i o t i

irrespective of the method by which the solution is obtained. A solotion obt:lined even by intelligent guessing is the only correct sollrtion. The import:lnce of this theorem will be appreciated when we discuss-the method of ima; ~ e s in Section 4-1.

To prove the uniqueness theorem. suppose a volume r i s bo~inded outside by a surface So, which may be a surface at infinity. Inside the closed surhce So there are a number of charged conducting bodies with surfaces S,, S,, . . . ,Sn at specified potentials, as depicted in the two-dimensional Fig. 4-2. Now :lssumr th:ll. contrnry to the uniqueness theorem, there are two solutions, V, and V,, to Poisson's equation in 5 :

140 SOLUTION OF ELECTGO7F-TIC PROBLEMS I 4

..-, \

Also assume that both Vl and V, satisfy the same boundary conditions on S , , S,, . . . , S, and So. Let us try to define a new difference potential

From Eqs. (4L29a) and (4-29b), we see that V, satisfies Laplace's equation in s

O n conducting boundaries the potentials are specified and I/, = 0. Recalling the vector identity (Problem 2-18),

V . (,/'A) = ,/'V . A + I\ . Vj'. (4-32)

and letting f = V, and A = VV,, we have

where, because of Eq. (4-31). the first term on the right side vanishes. Integration of Eq. (4 -33) ovcs n vnlumc r yiclds ---.

where a, denotes the unit normal outward from r. Surface S consists of So as well as S,, S,, . . . , and S,. Over the conducting boundaries, v, = 0. Over the large surface So, which encloses the whole system, the surface integral on the left side of Eq. (4-34) can be evaluated by considering So as the surface of a very large sphere with radius R. As R increases, both V, and V, (and therefore also I/,) fall off as 1/R; consequently. VV, falls off as 1/R2, making the integrand (5 VV,) fall off as 1/R3. The surfacc area So, however, increases as R2. Hence the surface integral on the left side of Eq. (4-34) decreases as 1/R and approaches zero at infinity. So must also the volume integral on the right side. We have

Jr jVv,12 do = 0 . (4-35)

Since the integrand IVV,I2 is nonnegative everywhere, Eq. (4-35) can be satisfied only if IVI/,I is identically zero. A vanishing gradient everywhere means that T/, has the same value at all points in z as it has on the bounding surfaces, S,, S ,,.. . . , S,, where V, = 0. It follows that I/, = 0 throughout the volume z. Therefore V, = V2, and there is only one possible solution.

It is easy to see that the uniqueness theorem holds if the surface charge distributions ( p , = E E , = -E JV/Jn), rather than thc potentials, of the conducting bodics

. . are specified. In such a case, VV, will be zero, which in turn, makes the left side of Eq. (4 -34) vanish and leads to thc same conclusion. In fact, thc uniqucncss thcorcrn applies even if an inhomogeneous dielectric (one whose permittivity varies with position) is present. The proof, however, is more involved and will be omitted here.

,IS [cell as ge burface Eq. (4--34) i radius R. sequently, rfa& area Eq. (4-34) le integral

e satisfied hat V , has ( 2 : . . . , s,,, e V, = V,,

eft SL.. of ;s theorem x ies with ~itted here.

L ' F i 4-4 1 METHOD OF IMAGES 141

- 1 $- ! I t

4-4 METHOD OF I M A ~ E S b

. . I \ * i -

There is a cladi(o[%l~ctro~tatic problems with boundary conditions that appear to be difficult to satis? if the governing Laplace's quation is to be solved directly, but the conditions oh the bounding surfaces in these can be set up by appropriate image (equiva1ent)kharges and the potential distributions can then be determined in a straightforward danner. This method of replac{hg bounding surfaces by appropriate image charges in lieu of a formal solution of Laplace's equation is called the method

, of images. .

, Consider the case of a positive point char&, Q, located at a distance d above a . large grounded (zero-potential) conduc~ing plane, as shown in Fig. 4-3(a). The problem is to find thk. potential at every point above the conducting plane ( y > 0). 'I Ilc formal proccdurc for doing 50 would b~ LO sdvc Ldplacc's cq~~ut lon in C:rrtca~,m . coordmates

which must hold for i, > 0 except at the point charge. The solution V ( s , J., z ) ahould satisfy the following conditions:

1. At all points on the grounded conducting plane, the potential is zero: that is,

2. At points very close to Q, the potential-approaches that of the pomi charge alone; that is

0 v-+- ~ ? C E ~ R '

where R is the distance to Q.

3. At points very far from Q(x - j CQ, y - + cc, or z - & a), the potential approaches zero.

Grounded plane conductor

- - - -\ (a) Physical arrangement.

- 0 (Image charge) . I (b) Image charge and field lines.

Fig. 4-3 Point charge ;rnd grounded plilne condt~ctor.

142 SOLUTION OF ELECTROSTATIC PROBLEMS / 4 c . I * .

i 4. The potential function is even with respect to the x and z coordinates; that is,

It does appear difficult to construct a solution for V that will satisfy all of these conditions.

From another point of view, we' may reason that the presence of a positive ; charge Q at y = d would induce negative charges on the surface of the conducting ; plane, resulting in a surface charge density p,. Hence the potential at points above : the conducting plane would be #

where R , is the distance from ds to the point under consideration and S is the surface of the entire conducting plane. The trouble here is that p , must first be determined fro111 tlic houndary condilion I.(\. . 0. :) =: 0. Mmcovcr, the indic~tcd surface intcp,il is dillicult to cvaluatc cvcn after p , ha> bccn tictcrmlncd at cvcry point on tlic conducting plane. In the following subsections, we demonstrate how the method of images greatly simplifies these problems.

4-4.1 Point Charge and Conducting Planes

The problem in Fig. 4-3(a) is that of a positive point charge, Q, located at a distance d above a large plane conductor that is at zero potential. If we remove the conductor and replace it by an image point charge - Q at y = - d, then the potential at a point P(x, y, z) in the y > 0 region is

where

R+ = [x2 + ( y - d ) 2 + z 2 ] l i Z , I

R- = [ x 2 + ( ~ + d ) ~ + z 2 ] l i 2 .

It is easy to prove by direct substitution (Problem P.4-5a) that V(x , y, z ) in Eq. (4-37) satisfies the Laplace's equation in Eq. (4-36). and it is obvious that all four conditions listed after Eq. (4-36) arc satislicd. Thcrcforc Eq. (4-37) is a solution of this problem; and, in view of the uniqueness theorem, it is the only solution.

Electric field intensity E in the y > 0 region can be found easily from - V V with . Eq. (4-37). It is exactly the same as that between two point charges, i- Q and - 0,

i - .- . !

$; that is,': ,' . 8 2

/ '

1 of these

1 pos~tive mducting nts above,

ie surface Termined e Integral tb-3n-

let,. of

I r k l ~ ~ c e onductor .t a point

(4-37)

q. (4-37) ) n + n ? s 3rotr&.

<

vv with illd -Q,

I

I I

I / / - I

1---- p----• I +Q 1 4-Q +t? -Q

(a) Physical arrangemeht. (b) Equivalent image-charge (c) Forces on charge Q. arrangement.

- Fig. 4-4 Point chard and perpendicular conduct& planes.

spaced a distance ?d apart. A few of the field lines are shown in Fig. 4-3(b) The solution of this elec ostatic problem by the method of images is extremely simple; but it must be asized that the image charge is located ourside the region in which the field is to be determined. In this problem the point charges + Q and - Q cutmor bu used to ca cula~c the V or E in the y < 0 region. As P matter of fact, both V and E are zero in he y < 0 region. Can you bxplain that? t

It is readily seed that the electric field of a line charge p, above nn infinite conducting plane can be found from p, ;ind its imiigc - p , (with tile condticting plane removed).

: I - 3 A pusilric p o i ~ ~ l d1i11.g~ Q is I O C : L L C ~ .LL d~hlii l lcc~ d l 1111d i f i , rei- pectivel~, from two grounded perpendicular conducting half-planes, as shown in Fig. 4-4(a). ~ e t e r m i h e the force on Q caused by the charges induced on the planes.

Solution: A formal solution of Poisson's equation, subject to the zero-potential boundary condition at the conducting half-plahes, would be quite difficult. Now an image charge - Q in the fourth quadrant would make the potential of the horizontal half-plane (but not that of the vertical half-plane) zero. Similarly, an image charge - Q in the second uadrant would make the potential of the vertical half-plane 2 (but not that of the brimntal plane) zero. But if a third image charge + Q is added in the third quadrant, we see from symmetry that the image-charge arrangement in Fig. 4-4(b) satisfies the zero-potential boundary condition on both half-planes and is electrically-eqyivaknt to the physical arrangement in Fig. 4-4(a).

Negative surface charges will be induced on the half-planes, but their effect on Q can be determined from that of the three image charges. Referring to Fig. 4-4(c), we have, for the net force on Q,

1144 SOLUTION OF El ';tTRCJ$TATIC PROBLEMS 1 4

where

F, = Q - 4ns,(~d,)~ '

F2 = -ax Q2 4 ~ 6 ~ ( 2 d ~ ) ~ '

F, = Q2 a 2d1 + a,2d2). I 4rrc0[(2d,)' + (2d2)2]31' ( "

Therefore,

Q2 F=- d2 '

The electric potential and electjic field intensity at points in the first yuodt-at~t and the surface charge density induced on the two half-planes can also be found from the system of four charges.

- 1-4.2 Line Charge and Parallel 1. Conducting Cylinder

We now consider the problem of a line charge p, (Clm) located at a distance d from the axis of a parallel, conducting, circular cylinder of radius a. Both the line charge and the conducting cylinder are assumed to be infinitely long. Figure 4-5(a) shows a cross section of this arrangement. Preparatory to the solution of this problem by the method of images, we note the following: (1) The imagcmust be a parallel line charge inside the cylinder in order to make the cylindrical surface at r = u an equipotentiai surface. Let us call this image line charge pi. (2) Because of symmetry with respect to the line OP, the image line charge must lie somewhere along OP, say at point P,, which is at a distance di from the axis (Fig. 4-5b). We need to determine the two unknowns, pi and d,.

,--I

/ I

\ \

(a) Line charge and Parallel conducting cylinder. (b) Line charge and its Image.

Fig. 4-5 Cross section of line charge and its image in a parallel conducting circular cylinder.

c l r ~ f and nd from

: d from : charge shows a n by thc c chargc mtcntial respect

?oint Pi,

. ,:

. . i 4-4 1 METHOD OF IMAGES 145 ,

, i ! r :, . * * * .i - 8 .- . : : let u$ assume that i

I "i (4-38)

At t h i ~ stage, Fq. (4-181 is just a trial solution (an intelligent guess), and we are not sure that it will hold tbue. We will, on the one hand, proceed with this trial solution until we find that it $ils to satisfy the boundary conditions. On the other hand, if Eq. (4-38) leads to a iolutian that does sitisfy dl1 boundary conditions, then by the uniqueness theored 14 is the only solurion. Our next job will be to see whether we can determine d;. . 3

The electric poiedtial at a distance r from a line charge of density p, can be obtained by integrating the electric field intensity E given in Eq. (3-36).

Note that the refeihce point for zero potential, r,, cannot be at infinity becais: . setting ro = in ~ q . ( 4 - 3 9 ) would make V infinite everywhere else. Let ui leave

r0 unspecified for the time being. The potential at a point on or outside the cylindricnl surface is obtained by addin: the contributions ofp, and pi. In particular, at a point .I on the cylindrical surface shown in Fig. 4-5(b),.we have

In Eq. (4-40) we have chosen, for simplicity, a point equidistant from p, and p, as the reference point for zero potential so that the In ro terms cancel. Otherwise. s constant term should be included in the right side of Eq. (4-40), but it would not affect what follows. Equipotential surfaces are specified by

r i - = Constant. r (4-41)

If an equipotential surface is to coincide with the cylindrical surface (OX = , l ) , ihc point P, must be louted in such a way as to mate triangles OMP, and OPM s~mllar. Note that t-two triangles already have one common angle, L MOP,. Point P, should be chosen70 make ,L OMP, = L: OPM. We have

rt di a . - =-- --- d - Constant. r q

, . * i 1 t

146 SOLUTION OF ELECTROSTATIC PROBLEMS I 4 ,! A % I r , ) I r . t

From Eq. (4-42) we see that if

the image line charge -p,, together with p,, will make the dashed cylindrical surface in Fig. 4-5(b) equipotential. As the point M changes its location on the dashed circle, both ri and r will change; but their ratio remains a constant that equals ald. Point Pi is called the inverse point of P with respect to a circle of radius a.

The image line charge -p, can thcn replace the cylindrical conducting surface. and V and E at any point outside the surface can be dcter~nincd from thc line charges p, and -p,. By symmetry, we find that the parallel cylindrical surface surrounding the original line charge p, withadius a and its axis at a distance 0, to the right of P is also an equipotential surface. This observation enables us to calculate the capacitance per unit length of an open-wire transmission line consisting of two parallel conductors of circular cross section.

-1. Example 4-4 Determine the capacitance per unit length between two long, parallel, circular conducting wires of radius a. The axes of the wires are separated by a distance D.

Solution: Refer to the cross section of the two-wire transmission line shown in Fig. 4 6. Thc cquipotcnti:ll surfxcs of the two wircs can bc considcrcd to hnvc been generated by a pair of line charges p, and,-p, separated by a distance (D - 2 4 ) = d - di. The potential difference between the two wires is that between any two points on their respective wires. Let subscripts 1 and 2 denote the wires surrounding the equivalent line charges p, and -p, respectively. We have, from Eqs. (4-40) and (4-421,

Fig. 4-6 Cross section of two-wire transmission line and equivalent line charges (Example 4-4).

t

(4 -43)

11 surface led circle, ~ / d . Point

g surface, ie charges rounding right df I' le capaci- o parallel

P

J , p . ~ . llel, ;?WC a

,hewn in have been

3 - ?dl) = two points unding the 4-40) arld

0

1

4 1 <r.- - - ' and, similarly, . - 1 .

rc a I

t V, = -L in

2nco , - . i ' - 1'

Hence the capacitarib per unit length is

d = f (n + V r ~ 2 - 4 2 ) .

Using Eq. (4-45) in kq. (4 -44). y c havc

Since In [x + fiZq] = cash-I x

for x 1 1, Eq. (4-46) can be written alternativefy as

4-4.3 Point Charge dnd Cbnducting Sphere .

t The method ok images can also be applied to solve the electrosrat~c problem of a point charge in the,prerence of a spherical conductor. Referring to Flg. 4-7(a) where a positive poidt charge Q is located at a distanced from the center ofa grounded conducting spher6-okradius a (a < d), we now proceed to find the V and E at polnts external to the sphere. By reason of symmetry, we expect the image charge Q, to he a negative poi6t Ch:lrpu .sitwitrcl iwidc I I I C S ~ I I C I C :111d 011 iiw lint joltling 0 ,111i1

Q. Let i t be at a dialunce di l i m 0. I t is obvious that Ql cannot bc equal to -0, since - Q a>dbthe oribinal Q do not make the spherical surface R = a a zero-potential surface as required. (Whd would the zero-potential surface be if Ql = -Q?) We must, therefore, treat both di and Q1 as unknowns.

' The other solution. d = i ( ~ - 4-1, is discarded because both D and d arc usually much larger than a.

L t , ,/--- , , .'

/' I.

I I \

i

v = o\ . - -

1 .

( a ) Point charge and grounded conducting sphere. (b) Point charge and its image.

Fig. 4-7 Point charge and its image in a grounded sphere.

In Fig. 4-7(b) the conducting sphere has been replaced by the image point cliiirgc Pi. wliicll makes thc po(e1itii11 :it ;dl points on the sphcric:ll s ~ i r f x c R = i l

zero. At a typical point M , the p)(csli:~l c;wed by Q and Qi is

which requires r i - - - --= r

Constant. Q

(4-49)

Noting that the requirement or. the ratio rJr is the same as that in Eq. (4-41). we conclude from Eqs. (4-42), (4-43), and (4-49) that

and

, The point Q, is, thus, the inverse point of Q with respect to a circle of radius a. The V and E of all points external to the grounded sphere can now be calculated from the V and E caused by the two point charges Q and - aQ/d.

- A " Example 4-5 A point charge Q is at a distance d from the center of a grounded conducting sphere of radius a (a < d). Determine (a) the charge distribution induced on the surface of the sphere, and (b) the total charge induced on the sphere.

nage. , , -

agc point 'acc R = a

u n h e d frolr 'he I

grounded 'n induced

4 Solution: he-bhysicb problem is that shown id Fig. 4-7(a). We solve the problem '

by the method of imagks and replace the grounded sphere by the image charge Q, =

-uQ/d at a d i s h c c d = u2/d ldrom the center of the sphcre, as shown in Fig. 4-8. The electric potential 1 at an arbitrary point P(R, 0) is

Q V(R, 0) = - - - - 4 ~ 6 0 (iQ , d i Q ) '

where, by the law of cobines,

RQ = [ R 2 + d 2 - 2Rd cos 81'1' and (4-52a)

Using Eq. (4-52) in Eq. (4-53), we have

ER(R, 8) L- I< - J cos 0

+ d 2 - 2Rd cod 8)3n

- a [ R - (a3/d) cos 81 d [ R 2 + (a2/d)' - 2R(a2/d) cos 8]312 1. (3-54)

: i a) In order t - f i n d the'induced surface charge od the sphere, we set R = o in Eq. (4-54) and evaluate

1 . , s , Ps = € o E ~ ( a , 6 ) 9 (4-55) which yields the following aftet simplificat!on:

' e ( d 2 - nd) PJ = - 4na(s2 + d 2 - 2ad cos 8)3/2' (3-56)

. b) The total charge induced on the sphere is obtained by integrating p, over the surface of the sphere. We have

Total induced charge = $ p ds - - So2" J: p g 2 sin 0 do d$

a = -- Q = Qi . - % I' ,I

d ,:$ . . (4-57) . \'., .

. - 2 . ,

We note that the total induced charge is exactlj equalto the image charge Qi that replaced the sphere. Can you explain this?

If the conducting sphere is'electrically neutral and is not grounded, the image of a point charge Q at a distance d from the center of the sphere would still be Qi at di given, respectively, by Eqs. (4-50) and (4-51) in order to make the spherical surface R = u equipokntial. However, an additional point charge

at the center would be needed to make the net charge on the replaced sphere zero. The electrostatic problem of a point charge Q in the presence of an electrically neutral sphere can then be solved as a problem with three point charges: Q' at R = 0, Qi at R = a2/d, and Q at R = d.

4-5 BOUNDARY-VALUE PROBLEMS IN CARTESIAN COORDINATES

We have seen in the preceding section that the method of images is very useful in solving certain types of electrostatic problems involving free charges near conducting boundaries that are geometrically simple. However, if the problem consists of a system of conductors maintained at specified potentials and with no free charges, it cannot be solved by the method of images. This type of problem requires the solution of Laplace's equation. Example 4-1 was such a problem where the electric potential was a function of only one coordinate. Of course, Laplace's equation applied to three dimensions is a partial differential equation, where the potential is, in general, a function of all three coordinates. We will now develop a method for solving three- dimensional problems where the boundaries, over which the potential or its normal derivative is specified, coincide with the coordinate surfaces of an orthogonal, curvilinear coordinate system. In such cases the solution can be expressed as a product of three one-dimensional functions, each depending separately on one coordinate variable only. The procedure is called the method of separation of vuriubles.

" (4-57) .;. 3 :

charge QL ' - . i: I .

Ime zero. lectrically ' a t R = 0,

I

i useful in onducting I

lslsts of a charged, it le solution : potential :d to three ral, a func-

its ingEi . ~nal, c- L l i -

I prodlr-of

Laplace's equatibn for scalar electric potedlial V in Cartesian coordinates is '

aZv a2v azv - , + 7 + - - = 0 . ax- ay- ai2 (3-58)

To apply the method of separation-of variables, we assume that the solution V ( s , y, r) can be expressed as d product in the following form:

1

where Xb) , Y(Y), a& Z(Z) are functions, respectively, of x, y, and z qnly Substituting Eq. (4-59) in Eq. (498), we have .I

VVIIILU, wnen alvlaea, through by the product X(x)Y(y)Z(z), yields

Note that each of@hthree terms on k? i;: rid4 of Eq. (4-60) is a function of only one coordinate variabie And that only ordine7 ierjvztives are involved. In order for Eq. (4-60) to be satis@dFr a11 values of x, y, z, eacd 0: the three terms must be a constant. For instance, if wiidtRersntiate Eq. (4-60) with hspect to x, we have

?

since the other typ tkrms are independent of x. lati ti on (4-61) requires that . - 1

.-. .I * 1 1 d2X(x) X ( x ) d x l = - k:, (4-62)

,. ,3?$ +'*&\:.>: 3

,, . where k: is a constant of integration to be detehined fib the boundary conditions . of the problem. The negative sign on the right side of Eq. (4-62) is arbitrary, just as

the square sign on k, is arbitrary. The separation constant k, can be a real or an imaginary1 number.'lf k, is imaginary, k j is a negative real number, making - k,2 a positive real number. It is convenient to rewrite Eq. (4-62) as

and

- where the separation constants k , and k , will, in different from k,; but, because of Eq. (4-60), thc following condition must be satisfied:

Our problem has now been reduced to finding the appropriate solutions-X(x), Y ( y ) , and Z(z)-from the second-order ordinary differential equations, respectively, Eqs. (4-63), (4-64), and (4-65). The possible solutions of Eq. (4-63) are known from our study of ordinary differential equations with constant coefficients. They are listed in Table 4-1. That the listed solutions satisfy Eq. (4-63) is easily verified by direct substitution. The specified boundary conditions will determine the choice of the proper form of the solution and of the constanrs A and B or C and D. The solutions of Eqs. (4-64) and (4-65) for Y ( y ) and Z(z) are entirely similar.

Table 4-1 Possible Solutions of X"(x) + k f X ( x ) = 0

k f 2, x ( x ) Exponential formst of X ( x )

+ k A , sin kx + B, cos k x C,elk + Die-Jk - jk A , sinh k x + B2 cosh kx .' C2@ + D2e-L

The exponential forms of X ( x ) are related to the trigonometric and hyperbolic forms listed in the third column by the following formulas:

e ih = cosh kx f sinh kx, cosh kx = f (ek + e-"), sinh kx = f (d" - e-'").

I . ;-

.r (4-64)

(4-65) - n L: ~ t ,

(4-66)

1s X(x), ,pect~vely, ,own from They ,are

wlfied by a choice of :solutions

n* , ,

i Example 4-6 Two grounded, semi-infinite, parallel-plane electrodes are sepamted '

Oy ;I dihtancc h. A lh rd clectlodc pcrpcndiculqr to both is mamtalned at a uunLtant potential Vo (see Fig. 4-9). Determine the potential distribution in the region enclosed by the electrodes.

1

Solution: Referring, to the coordinates in Fig. 4-9, we write down the boundary conditions for the tentlal function V(x, y, z) as follows. 9

With V independent of z : !b v (x , Y, 4 = v(x , Y). . (4-67a)

In the x-dircctioi: V(O, Y ) = Vo (4-67b)

t V(cf4 Y) =3 0. (4-67c) In the y-directiog:

F V ( S , 0) = 0 (4-67d) 1 - 1 V(x, b) = 0. (4-67e)

Condition (4-67a) idpliea k, = 0 and, from Table 4-1, I Z(Z) = B,. (4-68)

The constant A. vanlshes because Z is independent of z. From Eq. (4-66), we have

2 I I

ky = - k 2 = k2 , (4-69)

where k is a real nudber. This choice of k implies that k , is imaginary and that ky is real. The-uwpf kt = jk, together with the condition of Eq. (4-67c), requires us to choose the exponehtialiy decreasing form for X ( s ) , which is

I X ( x ) = ~ , e - ~ . (4-70)

In the y-directio& k, = k. Condition (4-67d) indicates that the proper cholce for Y( y) from Table 4-1 is .

I Y,(y) = A , sin ky. (4-71)

Combining the solutions given by Eqs. (4-68), (4-70), and (4-71) in Eq. (4-59), we obtain

Vn(x, y) = ( B , D , A , ) ~ - ~ sin ky - - = Cne-k" sin k y , .+, (4-72)

where the arbitrary constant C, has been written for the pr~duct B,D,A,. Now, of the five boundary conditions listed in Eqs. (4-67a) through (4-67e),

we have used conditions (4-67a), (4-67c), and (4-67d). In order to meet condition (4-67e), we require

K(x, 6) = Cne-kx sin kb = 0,

which can be satisfied, for all values of x, only if

, sin kb = 0

kb = nx

1111 I< =-, 11 = 1 , 2 , 3 , . , ,. .*% + (4 -74) h

Therefore, Eq. (4-72) becomes nx Vn(x, y) = Cne-nnx'b sin - y. (4-75) .. b

Question: Why are 0 and negative integral values of n not included in Eq. (4-74)? We can readily verify by direct substitution that K(x, y) in Eq. (4-75) satisfies

the Laplace's equation (4-58). However, Vn(x, y) alone cannot satisfy remaining boundary condition (4-67b) at x = 0 for all values of y from 0 to b. Using the technique of expanding an arbitrary function within a specified interval into a Fourier series, we form the infinite sum

In order to evaluate the coefficients C,, we multiply both sides of Eq. (4-76) by mn

sin - y and integrate the products from y = 0 to y = 6: b

nx mn S: mx 2 Ji c,, sin - y sin - y dy = V, sin - y dy. (4-77)

n = 1 b . b b

The integral on the right side of Eq. (4-77) is easily evaluated:

mn if m is odd J,b V, sin - y dy = b

if m is even

(4-72)

:h (4-67e), t condition

(4-73)

f3 -74) r

(4-75)

q. (4-74)? 5 ) satisfies rctnaming technique rier series,

(4-76)

(4-76) by

P 7 7 )

,

(4-78)

. L

Each-integral on the ltft side of Eq. (4-77) is I, f

nn kmn C, b ({-m)x ~ J ~ ~ , , s i n ~ ~ s ~ b Y d y = - J 2 0 [COs y - cos

.I b b ' - 7 f m = n

, I 'r I (4-79) If rn + n.

Substituting Eqs. (bk) and (4-79) in Eq. (4-7$), we obtain

v (4-80)

i f n is w e n .

The desired ~otential distribution is, then, a suberposition of V,(x, y) in Eq. 14-75),'

cc

I..(s, y~ = 7 c,,~ - m m sin - y J ,IL 1

h

Equation (4-81) is a rather complicated expression to plot In two dimensions. hut. since thc amplit dc of tho sinc tcrms in the series dccrcares very rapldly as increases, only the fir t fe v terms are needed to obtain a good approxlmation. Several k equipotential lines a t skztched in Fig. 4-9. 1 Example 4-7 Con der the region enclosed on three sides by the grounded conducting planes show , in Fig. 4-10. The end plate on the left has a constant potential f Vo. All planes are as8unxd to be infinite in extent in the r-direction. Determine the potential distributioi within this region.

Solution: The boundary conditions for the potential function V(x , y. z ) are as follows. , 1

With V independent of 2 : -1- V(s, y, 2) = tqu, 4').

In the s-directioh:

Since Laplace's equatioa is a limar partial di8erential equatron, the supcrpos~tron of soluirons IS also r solution.

156 SOLUTION OF ELECTROSTP: IC PRQBLEMS 1 4 ' '

Fig. 4-10 Cross-sectional figure for Example 4-7. ,

In the y-direction: V(X, 0) = 0 (4-82d)

V(x, b) = 0. (4-82e) r

Condition (4-82a) implies k, = 0 and, from Table 4-1,

As a conscqucncc, Ey. (4-66) rcduccs to .. ... -- which is tlw same as Eq. (4-69) in Examplc 4-6.

The boundary conditions in the y-direction, Eqs. (4-82d) and Eq. (4-82e), are the same as those specified by Eqs. (4-6713) and (4-67e). To make V(x, 0) = 0 for all values of x between 0 and a, Y(0) must be zero, and we have

Y(y) = A , sin ky, (4-85)

as in Eq. (4-71). However, X(x) given by Eq. (4-70) is obviously not a solution here, because it does not satisfy the boundary condition (4-82c). In this case, it is convenient to use the general form for k, = jk given in the third column af Table 4-1. (The exponential solution form given in the last column could be used as well, but it

* would not be as convenient because it is not as easy to see the condition under which the sum of two exponehtial terms vanishes at x = a as it is to make a sinh term zero. This will be clear presently.) We have

X(x) = A , sinh kx + B, cosh kx. (4-86)

A relation exists between the arbitrary constants A , and B, because of the boundary condition in Eq. (4-82c), which demands that X(a) = 0; that is,

0 = A, sinh ka + B, cosh ka

sinh ka B2 = -A,.------.

cosh ka

--Ue), are' I) == O for

(4 -85)

tion here, onvenient 4-1.-(The ell, but it ion under sin11 term

(4-86)

boundary n

, . I- 4v0 i f n is odd G7:, = nn sinh (nnulb)'

4 A , sinh X(s - t r ) , 1. (4-87) i i

where A, has been krllten fbr A,/cosh ka. It is bvident that Eq. (4-87) satisfies the condition X(a) = 0, experience, we shouldf,be able to write the solution given in Eq. (4;87) directly, the steps leading 1 b it, as only a shift in the argument of the sinh functibn is make it vanish tit x = a.

Collecting Eqs. (4483), (4-85) and (4-87), we obtain the product solutlon

V,(x, y) = BoA,A3 sinh k (x - a) sirl ky

hn nn,, * Ck sinh - (x - a) sin - y, n = 1 , 2, 3, . . . , b b

(4-88)

where C. = B,A,A,, Bnd k has been set to equdl nn/b in order to satlsfy boundary condition (4- 82e). i

. We have now urbd all $the boundary condiiions except Eq. (4-82b), whlch may be satisfied by a ~odrierdseries expansion of v(0, y) = Vo over the interkal from y = 0 to y = b. We hdbe

I t 1177 11n <, = Y,((), y ) = -.- 1 C.:, sinh - o sin - - y , 0 < v < h . (4-89)

11 I> . 11 T- I ' 11 I

We note that Ecji (Hb) is of the same form as Eq. (4-76), except that C, is replaced by - Cn sinh (nna/b). *he values for thr, coefficieht C i can then be written down from Eq. (4-80).

I 0, if 11 is cvcn.

The desired potentihl istribut~oa v*;;Z:,s ;Z: enclosed region in Fig. 4- 10 is a summation of K(x, y) in Eq.

158 SOLUTION OF ELECTROSTATIC PROBLGMS / 4 I

s - , . .'<, ' * . , 4-6 BOUNDARY-VALUE PROBLEMS IN

CYLINDRICAL COORDINATES

For problems, with circular cylindrical boundaries we write the governing equations in the cylindrical coordinate system. Laplace's equation for scalar electric potential V in cylindrical coordinates is, from Eq. (4-8),

A general solution of Eq. (4-92) requires the knowledge of Bessel functions, which we do not discuss in this textbook: In situations where the lengthwise dimension of the cylindrical geometry is large compared to its radius, the associated field quantities may be considered to be appro5imately independent of z. In such cases, d2V/Sz' = 0 and Eq. (4-92) becomes a two-dimensional equation:

--- -r Applying the method of separation of variables, we assume a product solution

where R(r) and @(4) are, respectively, functions of r and 4 only. Substituting solution (4-94) in Eq. (4-93) and dividing by R(r)(D(+), we have

In Eq. (4-95) the first term on the left side is a function of r only, and the second term is a function of 4 only. (Note that ordinary derivatives have replaced partial derivatives.) For Eq. (4-95) to hold for all values of r and 4, each term must be a constant and be the negative of the other. We have

where k is a separation constant. Equation (4-97) can be rewritten as

dZ@& /

+ kz@(4) = 0. ' d42

(4-98) 1 ' .

This is of the same form as Eq. (4-63), and its solution can be any one of those listed in Table 4-1. For circular cylindrical configurations, potential functions and therefore

potential

(4-92)

which we on of the ' luantities Vilz2 = 0

(4-93)

i o n P

(4 1 wlution

(4 - 95)

: second ;:y-r;

(4-96)

(4- 97)

r

(4-98)

se listed . lerefore

I 4-6 1 B O U N ~ ~ ~ Y ~ A L U E PROBLEMS IN ljYLIND&:7AL COORDINATES 159

1 . 1 ! . fi i:' ; ! i

t , @()) are periodtk id ) and the hyperbolic fhctiohs do not apply. In fact, if the range , '

o f ) is unrestricted, k bus t be an integer. Let k equal n. The appropriate solution is , 4

, @()) = A, sin nq5 + Bi m s n), (4- 99)

where A, and B, are d b i t q r y constants. We now turn our httention to Eq. (4-96), which can be rearranged as

1 'i , where idteger n has b k n written for k, implying a 2n-range for 6 . The solution of

Eq. (4-100) is t R(r) = A,P + B,r-". (4- 101).

This can bc vcrilicd S, direct substitution. Taking the product of the solut~ons in i (4-99) and (4-101), , e obtain a general solution of the .--independent Laplace's equation (4-93) for cikcular cylindrical regions with an unrestricted range for 4 :

K(r, k) = rn(A, sin ng + in cos n d )

, +r-"(A; sin n p -i Bi cos , I & ) , i~ # 0. (4-107)

Depending on the bobndary conditions, the complete solution of a problem may be a summation of the rms in Eq. (4-102). It is bseful to note that, when the region of interest includes e cylindrical axis where ). = 0;the terms contnlnlng the r-" factor cannot exist. the other hand, if the region of interest includes the point at

zero as r - r co. infinity. thc terms cohiaining thc P factor cannot exist, since the potential must be

When the potentktl is not a function of ), ii = 0 and Eq. (4-98) becomes 1.

I d2@(4;) -- dq5

- 0 , (4- 103)

'2 The general solution Of Eq. (4- 103) is 8(4) = A,@ + B,. If there is no circumferential variation, A, vanished,' and wt? have

8 ( 4 ) = B 0 , . k = 0 . (4- 104)

The equation for ~ ( r j also becomes simpler whkn k = 0. We obtain from Eq. (4-96)

, ' The term A& should d retained i f there is circumferential variation, such as in problems rnvolwng a wedge.

The product of Eqs. (4-104) and (4-106) gives a solution that is independent of either z o r 4 :

.V(r) = Cl In r + C,, (4- 107)

where the arbitrary constants t, and C2 are determined from boundary conditions.

Example 4-8 Consider a very long coaxial cable. The inner conductor has a radius a and is maintained at a potential Vo. The outer conducjor has an inner radius b and is grounded. Determine the potential distribution i g ~ e space between the conductors.

Solution: Figure 4-11 shows a cross section of the coaxial cable. We &ume no =-dependence and, by symmetry, also no +dependence (k = 0). Therefore, the electric potential is n function of r only and is givcn by Eq. (4- 107).

.

. The boundary conditions are

Substitution of Eqs. (4-108a) and (4-108b) in Eq. (4-107) leads to two relations:

. C l l n b + C 2 = 0 , (4- 109 a)

. Cl Ina + C2 = Vo. (4- 109 b)

Expressions C; and C2 are readily determined:

Therefore, the potential distribution in the space's I r I b is

i - Obviously, equipotential surfaces are coaxial cylindrical surfaces. ,

i-

, - I

of either

(4 107)

ilditions.

, radius rac'f' b

ween . rle

,sume ti0 ic electric

(4-108a)

(4- lO8b)

tions:

(4- 1 O9a)

(4- l o b )

/ I

n f

(4- 110)

6 p * 4-6 t OUND&~-Y-WUE PROBLEMS IN I ~ ~ ~ ~ E I ~ ~ ~ ~ ~ ~ COORDINATES 161

' 1 ; . . ?' . ' :!;<is <

. . tJ ,, - b

I . :

1 I

I

I , tl

Fig. 4-12 Cross section of split \ \

/

A' circular cylir~bcr and cqu~potcnt~:ll

'----/ i: lines (Example 4-9).

Example 4-9 An i%finirely long, thin, pnducting circular tube of rndiub b is split i n ~ w o halves. The er half is kept at a potential V =. V, and the lower half at V = -*Vo. Determine distribution both inside and outside the tube. -

Solution: A cross section of the split circular tkbe is shown in Fig. 4-12. Since rhe tube is assumed to b! infjnltely long, the potential is independent of 2 and the two- dimensional Laplace'b equation (4-93) applies. The boundary conditions are:

These conditions are plotted in Fig. 4-13. Obviously V(r, 4) is an odd function of 4. We shall determine. P(r, @i inside and outside the tube separately.

a) Inside the tube,

I I I . I 1 ! I o T 2n i +' I 3 I . L----vo d L- , Fig. 4113 Boundary condition for . Example 4-9.

Because this region includes r = 0, terms containing the r-" factor cannot exist. Moreover, since V(r, 4) is an odd function of 4, the appropriate form of solution

V,(r, 4) = A,? sin n 4 .

However, a single such term d ~ e s not satisfy the boundary conditions specified in Eq. (4- 11 1). We form a series solution

m . -

W . 9 4) = 1 K(rY 4 n = 1

= 2 A.? sin n 4 , (4- 1 13) I n = 1

and require that Eq. (4- 11 1) be satisfied at r = b. This amounts to ex'panding the i rectangular wave (period = 2x) , shown in Fig. 4-13, into a Fourier sine series.

1

The coefficients An can be found by the method illustrated in Example 4-6. As a matter of fact, because-we already have the result in Eq. (4-80), we can directly write

if n is odd 4-7 (4-115) 1 IN St

The potential distribution inside the tube is obtained by substituting Eq. (4-115) h I !'

in Eq. (4- 1 13).

b) Outside the tube,

- In this region, the potential must decrease,to zero as r -r m. Terms cbntaining the factor r" cannot exist, and the appropriate form of solution is

= 2 Bnr-" sin n 4 .

?it i r ! 1, t

- ;I' '

(4-113)

inding the inc scries.

(4-1 14) P

4-' , a in directly

(4-115)

Zq. (4- 1 15)

(4- 116)

tontaining

P

I '

(4-1 17)

not exist. r i

rso~utio? . : '1'. . , . i

i , (4-112)

a

specified 1 I

The coefficients $,, in Eq. (4-1 18) are analogous to A,, in Eq. (4-114). From Eq. . (4-115) we obtaih 'I 6

Y ' I - !, '4 ' .n

I if n is odd ' B d = (4-1 19)

1 0 7 if n is even.

Therefore, the pdtcntial distribution outside the tube is

Several equipotefitial lines both inside and outside the tube have been sketched in Fig. 4-12. 1

The general equation in spherical coordinates is a very involved discussion to cases where the electric potential is

independenteof the llzirnuthal angle 4. Even with this limitation we will need to introduce some new functions. From Eq. (4-.9) we have

Applying the method of separation of variables, we assume a product soiutlon

I' V(R, 0) = T(R)O(B). (4-112)

Substituti-)his Bolution in Eq. (4-121) yields, after rearrangement, I

In Eq. (4-123) the first term on the left side & a idfiction of R only, and the second term is a function of 0 onb. If the equation is to hold for all values of R and 0, each term

164 SOLUTION OF ELECTROSTATIC PROBLEMS 1 4

must be a constant and be the negative of the other. We write

- where k is a separation constant. We must now solve the two second-order, ordinary differential equations (4-124). and (4-125).

Equation (4-124) can be rewritten as

which has a solution of the form

r, ,(R) = AnRn + B,R-'"'I). - In Eq. (4-127). A,, and B,, are arbitrary constants. and the fitbwjng relation between 11 and k can be verified by substitution:

n ( / t + I ) = I?, ( 4 - 1 3 ) i i where n = 0, 1 , 2,.. . . is a positive integer. k

1 With the value of k2 given in Eq. (4-128), we have, from Eq. (4-125), !

+ n(n + I)@(@) sin 8 = 0, r

which is a form of Legendre's equation. For problems involving the full range of 0, I

t from 0 to n, the solutions to Legendre's equation (4-129) are called Legendre j~ncrions. usually denoted by P(cos 8). Since Legendre functions for integral values of n are t

E polynomials in cos 0, they are also called Legendre polynomiuls. We writc 1

@,(O) = P, (COS 0). (4-130)

Table 4-2 lists the expressibns for Legendre polynomialst for several values of n.

Combining solutions (4-127) and (4-130) in Eq. (4-122). we have, for spherical boundary-value problems with no azimuthal variation,

Depending on the boundary conditions of the pivcin problem, the completc solution may be a summation of the terms in Eq. (4-131). We illustrate the application of

' Actually Legendre polynomials are Legendre functions of the first kind. There is another set of solutions to Legendre's equation, called Legendre functions ofxhe second kind; but they have singularities at 0 = 0 and n and must, therefore, be excluded if the polar axis is a reglon of interest.

(4-125)

xdinary : . I

/ '

(4 - 120)

(4- 127)

b e t y n

(4-128)

(4-1 29)

nge of 0, fuitctions, s f n are

(4- 130)

ues of n.

spherical

( 4 ~ 1 )

:soil, . cation of

of solutions ies at 0 = 0

f P" (cos 0) 4 t . , , :

. ' ' \ " . , t 1 cos Q

2 A(3 cos2 0 - 1) 3 :: +(3 C O S ~ e - 3 cos 8)

Lcgcndre po~yno$ids in thc Solution of a sidple boundary-v;lluc problem in the ,

following example. . I

Example 4-10 An Uncharged conducting sphere of radius b is placed in an initially uniform electric field $, = azEo. Determine (a) the potential distribution V(R. 0 ) and (b) the electric field idtensity E(R. 0) after the introduction of the sphere.

Solution: After the conduct~ng sphere is introdsccd into the clectrlc lield, :l sep;lr:i- llon and redistributiqp of charges will take place in such a way that the surface of thc sphere is The electric field intensity within the sphere 1s

will intersect the surface normally, and the the sphere will not be affected appreclably.

The geometry of thistproblem is depicted in Fig. 4-14. The potentla1 is, obviously, independent of the adnutbal angle 4, and the solhtion obtained in this section applies.

k ' B.1

.

. Fig. 4-14 Conducting sphere electric field (Example 4-10).

in a uniform

a) To determine the potential distribution V(R, 6) for R 2 b, we note the following boundary conditions : .

Equation (4-132b) is a statement that the original Eo is not disturbed at points very far away from the sphere. By using Eq. (4-131), we write the general solution as

However, in view of Eq. (4-132b), all An except A , must vanish, and A , = - E,. We have, from Eq. (4-133) and Table 4-2,

-1 = B,R- ' + (B,R-' - E,R) cos f3 + BnR-(ntl)P,,(cos 8), R 2 b

n = 2

(4- 134)

Actually the first term on the right side of Eq. (4-134) corresponda to the potential of a charged sphere. Since the sphere is uncharged, Bo = 0, and Eq. (4-134) becomes

(4-13.5)

Now applying boundary condition (4-132a) at R = b, we require

0 = ( -2.- E 0 b ) cos 6 + 2 Bnb-'n+l)P,,(cos e), n = 2

from which we obtain B, = Eob3

and Bn=O, n 2 . 2 .

' For t h ~ s problem it is convenient to assume V = 0 in the equatorial plane (0 = 7~12). which leads to V(b .0 ) = 0, since the surface of the conducting, sphere is equipotential. (Sec Problem P.4-21 for

- V(b, O) = V,.) .

1 solution

potential 1. (4-134)

w h ~ h ldds P.4-21 for t

1

i 4 . , p. ,, - j i 1

, We hive, tli$aliy,;~rom'~q. (4-135), . , 0 4 . j

* r i ) ( R , ~ ) = - E , p R z ~ . (4-136) . ; i . #

, , b) The electric f i e l ~ intensity E(R, 0) for R'F b can be easily determined from -VV(R,8): 1 .

F E(R, 0) = aREi . + aoE,,

where ' (4-137a)

f ,: dV

I p and ! a = - a = ~ o [ l + 2 ( ' ~ ] c o s 0 , R k b (4-137b)

The surfac; charbe density on the sphere can be found by noting

which is p;oportional to cos 8, bemg zero at O = n/2. Some equipotent~al and - field lines are gketched in Fig. 4-14.

In this chapter wd'ha~e discussed the analytical solution of electrostatic problems by the method ~f imQes and by direct solution of Laplace's equation The method of images is useful when charges exist near conductlng bodies of a simplc and com- patible geometry: a point charge near a conducting sphere or an infinite conductlng plnnc; and it linc chafgc I G I ~ iL pat~ilcl wiiriucling cylindcr or .I p:~r:iIlcl conductitlg plmc. Tlic aolillioii a/' Laplacc'a cquauon by the method of separation of variables requires that the bouhdaries coincide with cooidinate surfaces. These requirements restrict the u se f~ ln~ss~o f both methods. In practical problems we arc often faced with more complicated bqpdarics, which are not amenable to neat analytical solutions. In such cases, we must resort to approximate grkphical or numerical methods. These methods are beyond the scope of this book.+

I . >

REVIEW QUESTIONS I

, !

R.4-1 Write Poisson's lequation in vector notation -1. j

a) for a sinipte b) for a linciw and isotropic, hut inl~omo~cncous ti~cdiunl.

R.4-2 ~epeablfi cartdsian coordinates both p&ts of R.4-1.

' See, for instance, B. D. Popovit. Introductory Engineering ~ k ~ t r o m a ~ n e t i c s , Addison-Waley Publishing Co. (1971), Chapter 5.

'

,; " R.4-3 Write Laplaa's equation for a simple medium t . > 3 . '; a) in vector notation, b) in Cartesian coordinates.

R.4-4 If V'U = 0, why does it not follow that U is identically zero?

R.4-5 A fixed voltage is connected across a parallel-plate capacitor.

a) Does the electric field intensity in the space between the plates depend on the permittivity of the medium?

P b) Does the electric flux density depend on the permittivity of the medium?

Explain. 1 i i'

I ' R.4-6 Assume that fixed charges +Q and - Q are deposited on the plates of an isolated parallel- I plate capacitor. ! '

a) Does the electrlc field intensity in the space between the plates depend on the permittlvlty of the medlum'? ,

b) Does the clectric flux density depend on the perm~ttivlty of the medium'? ! Explain. I

R.4-7 Why is the electrostatic potential continuous at a boundary? - R.4-8 State in words the uniqueness theorem of electrostat~cs. -\-

R.4-9 What is the image of a spherical cloud of electrons with respect to an infinite conducting : plane?

R.4-10 Why cannot the point at infinity be used as the point for the zero reference potentlal for 1 an infinite line charge as it is for a point charge? What is the physical reason for this difference?

I R.4-11 What is the image of an infinitely long line charge of density p, with respect to a parallel , conducting circular cylinder? i

R.4-12 Where is the zero-potential surface of the two-wire transmission line in Fig. 4-6? 1 I

R.4-13 In finding the surface charge induced on a grounded sphere by a point charge, can we 1 set R = a in Eq. (4-52) and then evaluate ps by -ro aV(a, QWR? Explain. i i

R.4-14 What is the method of separation of variables? Under what conditions is it useful in ,

solving Laplace's equation?

R.4-15 What are boundary-varue problems? I L

I

R.4-16 Can all three separation constants ( k x , k,, and k,) in Cartesian coordinates be real? Can they all be imaginary? Explain. i R.4-17 Can the separation constant k in the solution of the two-dimensional Laplace's equa- ! ,'7 tion (4-97) be imaginary? I

1 i ,

R.4-18 What should we d o to modify the solution in Eq. (4-'110) for Examplc 4-8 i f the inncr i , - i conductor of the coaxial cable is grounded and the outer conductor is kept a t a potential Vo?

. .. ' 1 ; z I . . .

R.4-19 What should we d o to modify the solution in Eq. (4-1 16) for Example 4-9 if the con- * . . , ducting circular cylinder is split vertically in two halves, with V = Vo for - n/2 c 4 n/2 and

V = - Vo for x/2 < 4 < 3n/2? i 1

"I ui parallel-

xrmittivlty

/-'

conductmg

otent~al for d~fference?

o a paallel

1-6?

rge, can we

it useful in

e real? Can

a t P q u a -

I

if the inner ntial Vo?

I if the cod: c lr/2 and

- ,,?+.*: - 2 "' < .. .. . ., . A b. r ; .: .".

' i

PROBLEMS 169 ,

,. i !; { I . .

R.4-20 Can functibns ' k ( ~ , = cos 0, where C, and C, are

, '? arbitrary constants, be db1utid.m of Explain.

, " 7 r 1 . .

4 . * PROBLEMS I ! ' [

P.4-1 The upper and lower conductmg plates of a k g e parallel-plate capacitor are seprrated by a distance d and mai tti in~d at potentials Vo and respct/vely. A dielectric slab of dielectric constant r, and u n i f o d hiekoess 0.8d is placed over the lower plate. Arsumlng negligble fringing t effect, determine , r

a) the potential adti eleqtric field distribution inihe dielectric slab, b) the potcntial aHd electric field distribution iH the air space between the dielectric slab

and the upper plate, C) the surface chdfge densities on the upper +d lower plates.

P.4-2 Prove that the scalar potential V in Eq. (3-56) satisfies Po~sson's equation; Eil. 14-6). . , ,

PA-3 Prove that a 60tential function satisfying Laplace's equation in a given region possesses no maximum or minimum within the region.

'

P.4-4 Verify that >

r V, = C , / R and V, = C2z/(.y2 + yZ + i 2 ) ' I 2 ,

where C, and C, are arbitrary constants, are solutions of Laplace's equation

P.4-5 Assume a point charge Q above an infinite cdnducting plane at J> = 0.

a) Prove that V(x,'): i) in Eq. (4-37) satisfies Laplace's equation ~f the conduct~ng plane is maintained at tero uotential.

b) What should t ~ k expression for V ( x , y;z) be i6the conducting plane has a nonzero potential V,'? 4

c) what-is ths e l~~t ros ta t i c force of attraction between the charge Q and the conducting plane?

.I: PA-6 Assume that spkce between the inner and outer conductors of a long coaxial cylindrical structure is filled with ad electron cloud having'a volume density of charge p = /I /r for a < r c b, where a and b are, respdctively, the radii of the inner and outer conductors. The inner conductor is maintained at a potential V,, and the outer conductor is grounded. Determine the potential distribution in the reg& a < r < b by solving Poisson's equation.

P.4-7 A point charge' Q exists a a distance d above a large grounded conducting plane. Determine

a) the surface chatke density p,, b) the lmal_qharge indqced on the condxiillg $erz. I

,+ P.4-8 Determine the &ystems of image chi\;ib (hat ;1111 replace the conducting boundaries that are maintained at Zero potential for

' a) a point charge located between two large, grounded, parallel conducting planes as shown in Fig. 4-15(a), +

b) an infinite line charge p; located midway between two large, intersecting conducting planes forming P 60-degree angle, as shown in Fig. 4-15(b).

3 . '< , , . . :.L'*.,f.,r . ., , *>,f..l .? , . , , ' : " , .: .~ . , . . , t j ~ ? . i i ~ + " ~ : ~ : ; , ~ E E 2 ~ \ ~ ~ ~ i j . 5 ~ !:;I;: ~ ;; .y.: " ' '

I .,i: * . ' ! ' . *::, .... . t : . , ,.,; <.,, i' . . : , " " - . ' , . ' , '

' . . . . , . , .,.. . . :, i z , x- : ; , . . , , , , .:;. . . , . .Y! . , ; , 4 :,,,.*.. - .&-,+".'-. ;,. . .-. ,.;:.. -..'.:.;.-i. ': ., I . . I ; . . . . ,. [ -ii ., . _ . ._ : \ 8 . % .. . . p . . >

, ! - , ' . ' : , ,, . . ,

, . . . . . . , a , . , 3 . .

' . I.. - ' . 170 , %OLU- '2.N OF ELECTROSTATIC PROBLEMS 1 4 , . . .

- (a) Point charge between

grounded parallel planes.

' ( 1

G a

Fig. 4-15 Diagrams - - for Problem P.4-8. (b) Line charge between

grounded intersecting planes.,

1 : P.4-9 Two infinitely long, parallel line charges with line densities pc. and - p , are located at

. - - b b & - + - and z = - - i f 2 2

respectively. Find the equations for the equipotential surfaces, and sketch a typical pair. i i

P.4-10 Determine the capacitance per unit length of a two-wire transmission line with parallel I

conducting cylinders of dillkrent radii u , and o,, tllcir axcs being scp;wutcd by n distancc D (where D > a, + a,).

i 1. -

PA-I 1 A straight conducting wire or radius tr is p;u;~llcl to and nt height 11 from thc surbce of I

the earth. Assuming that the earth is perfectly conducting, determine the capacitance per unit length between the wire and the earth. i P.4-12 A point charge Q is located inside and at distanced from the center of a grounded spherical conducting shell of radius b (where b > d). Use the method of images to determine

a) the potential distribution inside the shell, b) the charge density p, induced on the inncr surface of the shell.

P.4-13 Two dielectric media with dielectric constants E , and e2 are separated by a plane boundary at x = 0, as shown in Fig. 4-16. A point charge Q exists in medium 1 at distance d from the boundary.

a) Verify that the field in medium 1 can be obtained from Q and an image charge -Q,, both acting in medium 1.

(Image charge) I (Image charge)

Medium 2 (62) Medium 1 ( e l )

Fig. 4-16 Image charges in dielectric x = O media (Problem P.4- 13).

7f1ir.

r parall~l stance D

: bound- : d from

ge -QI:

* ,

. . ' .

Verify that the field in'medium 2 can be obtained $om Q and an image charge +Q,, both acting in mediud 2.

c) Determine Q, and Q,. [ ~ i n t : Consider neighboring boints P, and P , In media 1 and 2 , respectively and requird the continuity of the tangential component of the E-field and

of the normal coapon&t of the D-field.)

P.4-14 i n what way should we hodify the solution in E ~ . 14-91) for ~ x a r n ~ l e 4-7 rfthe boundary conditions on the top, bottbm, and rlght planes in Fig. 4-10 are dV/an = O?

PA-15 In what way shollld idmodi fy the solution in Eq. (4-91) lor Example 4-7 if the top, bottom, and left planes in F& 41-10 are grounded ( V = 0) and an end plate on the right is rnain- tained at a constant potential to?

P.4-16 Consider the rcctangulab region shown in Fig. 4.- 10 as thc cross scction of an cnclosurc formcd I~y'rour contlucli~~g pli~t~?i. I'hc Icl't mid right plntcs arc yroundcd, and the top and bottom pixto i ~ l r in;~imiiincd a1 constnht potentials v, and V, respstively. Determine the potential distribution insidc the enclosurq;

P.4-17 Consider a metallic rektrihgular box with sidcs a and b and height c. The side walls and thc bottom surface are grounded, The top surface is isolated and kept a t a constant potentla1 Vo. Determine the potential distribution inside the box.

PA-18 An infinitely long, thin, c nrluctinp circular cylinder of radius b is split in four quartcr- cylinders, as shown in Fig. 4-17. he quarter-cylinderi in the second ;~nd foilrtli L ~ I I I : \ i t < r \ l \ t < : \ S t

grounded. and those in the first and third ,l~iitdru~,[< 4rc i c j l ( ;I( i ~ o ~ c ~ t i a l s i; ; ~ s d - I; respcc- t i v d y , Dclcrn\iw t l ~ c pukalid ilbtvibutial both inr/de md outside the cylinder.

C ' .

Fig. 4-17 Cross section of long circular cylinder split in f o i r

I quarters (problem PA-18).

P.4-19 A long, grounded conductink cylinder of radius b is placed along the z-axis in an initially uniform electric field E, = a,E,. Deterpine potential distribution Y(r, (6) and clectric field in- 'tensity E(r, $) outside the cylinder.

'

P.4-20 A long dielectriccq4nder of tadius b and dielectric constaht 6, is placed along the z-axis in an initially uniform electric field ko = a,Eo. Determine V(r, #J) and E(r, 4) both inside and outside the dielectric cylinder.

P.4-21 Rework Example 4-10, asshmipg V(b, 0) = Vo in,Eq. (4-132a).

P.4-22 A dielectric sphere of radius b and dielectric constant 6, is placed in a n initially uniform electric field, Eo = aZEo, in air. Detmniqe V(R, B) and E(R, 0) both inside and outside the dielectric sphere.

-.I ';-;,...'.;>, , . , . * ' . , .. ,;, .. - 1 ."' '. ., ?...,.,;.,.*a:,> ') . 1 . ,

, , ' . * . ,-.> ,. .,! :, ,<-. -. -, , .,.

.,, , . 5.. ,,,'...,~,k-!,, a, .- , ,* . ,; .. :..,n~.w> 7 . c . r . d ,

! , . d 2 ,. . . .,; , . - , L .;.. ~.!'l . ' ; . /. I ' .. ;, ... ' , . . . . , ' , . , . .. , , ., ..,, ,*:,. , . :,, ; , 3 . ;: , .: >::-, y. , : z ,, i.!,,, ~ ;I!:, ;,: !y+y;::

, ' : , ' . . .-I . ,: , . . I ,:", , . <;i.;;p; . . . L,:, :',., ..; ,.; .r.: ,

, , .: .,, , . , . .

, , . 1 . . .< ,

5 / Steady Electric Currents . [ 1 :,. , . . . .

. , . , ,

5-1 INTRODUCTION i

In Chapters 3 and 4 we dealt with electrostatic problems, field problems associated i with electric charges at rest. We now consider the charges in motion that constitute \ current flow. There are several types of electric currents caused by the itloti011 of p e e i charges.+ Conduction currents in conductors and semiconductors are caused by drift motion of conduction electrons and/or holes; e l e c t r o l ~ ~ i i ~ ~ r c i ~ t s are the result of ! migration of positive and negative ions; and convection currents result from motion of clcctrons :rntl/or ions in a v:uxum. I n this chnpter wc shi~ll pay spccinl attention to 3 conduction currents that are governed by Ohm's law. We will proceed from the point form of Ohm's law that relates current density and electric field intensity and obtain the V = IR relationship in circuit theory. We will also introduce the concept

I principle of conservation of charge, we will show how to obtain a point relationship

I of electromotive force and derive the familiar Kirchhoff's voltage law. Using the i

between current and charge densities, a relationship called the equation of continuity 1 from which Kirchhoff's current law follows.

When a current flows across the interface between two media of different conductivities, certain boundary conditions must be satisfied, and the direction of

t

currcnr llow is ch:~ngcd. Wc will discu\s thcsc boundary contlitions. Wc will :11w i show that Sor a homogeneous conducting medium, the current density can be 1 expressed as the gradient of a scalar field, which satisfies Laplace's equation. Hence. j an analogous situation exists between steady-current and electrostatic fields that is the basis for mapping the potential distribution of an electrostatic problem in an electrolytic tank.

i 1 . The electrolyte in an electrolytic tank is essentially a liquid medium with a low

F conductivity, usually a diluted salt solution. Highly conducting metallic electrodes are inserted in the solution. When a voltage or potential difference is applied to the

j electrodes, an electric field is established within the solution, and the molecules of the electrolyte are decomposed into oppositely charged ions by a chemical process I

.,A , .. called electrolysis. Positive ions move in the direction of the electric field, and negative ! I

' In a time-varying situation, there is another type of current caused by bound charges. The time-rate of change of electric displacement leads to a displacement current. This will be discussed in Chapter 7.

t

e 5-2 1 CI ,nt. ' 1ENSIlY AND OHM'S LAW 173 . i

11

4

-. ) ions move in a directid opposite td the field, both contrib'uting to a current-flow in 1 *

, I the direction of tkb k fiel, . An experimental model ban be set up in an electrolytic tank, '

with electrodes dqpibpkr geometdcal shapes sirndating the boundaries in electrostatic

4 problems. The deasurkd potential distribution iii the electrolyte is then the solution to Laplace's eqlYation b r difficult-to-solve analytic problems having complex boundaries in a homogeneoys medium.

associated constitute iou of free ed by drift ie r e t of xn . 'on tier. .-.. to 1 frca the tensity and he w x e p t rJymg the

cld~mnship f conrinuity

3f different lirection of le will also jity can be ion. Hence, fields that

problem, in

I with a low c c l p y d e s p i $3 the noli of I: , process mi\ negative

'he rime-rate of 'hapter 7.

Convectioq. curreyts ark the result of the motion of positively or negatively charged particles in a jacuum or rarefied gas. Fahiliar examples are electron beams in a cathode-ray tube, >nd the violent motions of dharged particles in a thunderstorm. Convection curtents, t e result of hydrodynamic motion involving a mass transport, are not governed by Iinl's law. d The mechanism o conduction currents is diRerent from that of both electrolytic currents and convecttbn currehts. In their normal state, thc atoms of a conductor occupy regular positions in u crystalline structdre. The atoms consist of positively charged nuclei surrounded by electrons in a shell-like arrangement. The electrons in the inner shells are tightly bound to the nuclei and are not free to move away. The electrons in the outerrflost shells of a conductor atom do not completely fill the shells: they are valence or corlduction electrons, and are only very loosely bound to the nuclei. These latter electrons ,may wander from one atom to another in a random manner. The atoms, on the ~Gerage, remain electrically neutral, and there is no net drift motion of electrons. ,when an external~electric field i's applied on a conductor, an organized motion of the conduct.ion electrdns will result, producing an electric current. The average drift velocity of the clcctrdns i s vcry low (on the order of 10-' or lo-* m/s) even foi"vcry good conductors, because they collide with the atoms in the course of their qotion, dissipating part of their kinetic energy as heat. Even with the drift motioh of conduction electrons, a conductor remains electrically neutral. Electric forces prevent excess electrons from accumulating at any point in a conductor. We will sdow gnalytically that the charge density in a conductor decreases exponentially with tide. In a good conductor the charge density diminishes extremely rapidly toward zero as the state of equilibrium is approached.

1 5-2 CURRENT DENSITY *ND OHM'S LAW

Consider the sttady rhotion of one kind of charge carriers, each of charge q (which is negative for electrons), across an element of surface As with a velocity u, as shown in Fig. 5-1. If N is the h m b e r of charge carriers per unit volume, then in time At each

1 charge carrler moves a distance u At, and the amount of charge passing through the surface As is

AQ = N q u a, As At (C). . (5-1)

Since current is the tlme rate of change o f ~ h a r i e , we have

I . . i ' . P 174 STEADY ELECTRIC CURRENTS I 5

1 .

* : 1 I

Fig. 5-1 Conduction current due to drift motion of charge carriers across a surface. . I

\ 3

In Eq. (5-2), we have written As = a,As as a vector quantity. It is convenient to define a vector point function, yolurne current density, or simply current density, J, in amperes per square meter,

J = Nqu (A/m2); (5-3)

so that Eq. (5-2) can be written as - * . . -.

A1 = J . As. (5 -4)

The total current I flowing through an arbitrary surface S is then the flux of the J vector through S:

Noting that the product N q is in fact charge per unit volume, we may rewrite Eq. (5-3) as

which is the relation between the contiection current density and the velocity of the charge carrier.

In the case of conduction currents there may be more than one kind of charge carriers (electrons, holes, and ions) drifting with different velocities. Equation (5-3) should be generalized to read

As indicated in Section 5-1, conduction currents are the result of the drift motion of charge carriers under the influence of an applied electric field. The atoms remain neutral ( p = 0). It can be justified analytically that for most conducting materials

ivcnicnt to 2111 ciolsiry,

(5-3)

-4,

ax 01 me J

(5-5)

nay rewrite

(5 -6)

ocity of the

d 'of charge lation (5-3)

n - 7 )

:t motlon of 3ms remain ~g materials

, \ k~ DENSITY AND OHM'S LAW 175 4 '* i I ' I ,

the average drift directly propo;tioml to the electric field intensity. Con- sequently, we (5-3) or Eq. (5'7) i s

where the proportioba~ity coastant, o, is a mhcroscopic constitut~ve parameter of -

the medium called e nductiuity. Equation is a constitutive relation of the con-

11 ducting medium. Is ropld materials the linear relation Eq. (5-8) holds are called ohmic meqba. The unit for o is ampep per volt-meter (AP-m) , or siemens per meter (S/m;). Ca per, the most commonly used conductor, has a conductivity d 5.80 x lo7 (S/h). 0 n i h e other hand, hard rubber, a good insulator, has a conductivity o f only 10-I' IS/m). hppcnciix H 4 11\ts thc conduct~vlt~cs of some other frequently . uacd materials Howpver, note that, unlike the dielectric constant, the conductlvlty of materials valies dver an extremely wlde range. The reciprocal of conductlv~ty 1s called rrsi~tiuity, in oHrn meter (i2.m~. We prefer to use conductivity; there is really no compelling need to ube both conductlv~ty m d resistivity. .

We recall Ohm's iua from circuit theory that the voltage V,, across a resistance R, 111 which a current I flows from point 1 to point 2, is equal to RI; that is,

V12 = R I . (5-9)

Here R is usually a of conducting materid of a giyen lengh: V, is the voltage bctwecn two lcrn~inals I i11Id 2 ; i~nd I is the total current [lowing iron1 terminal 1 to tcrm~nal 2 through a frnitc cross 4cctlon.

Equation (5-9) 1s not a point reiatlon. Although there is little resemblance between Eq. (5-8) anti Eq.'(5-9), the former is gknerally referred to as the point form of Ohids law. it holds at all points in space, and D can be a function of space coordinates.

Let us'use the poiht form of Ohm's law to dirive the voltage-current relationship of a piece of homogWleous material of conductivity c, length L and uniform cross- section S, as shown iH Fig. 5-2. Within the conducting material, J = oE where both J and E are in the dirktion of current flow. The potential difference or voltage between

Y A

:'.,'.' " ' 8 . r , . <, ., . . , $> - t - .' terminals land 2 ist ' ' "' , . ' . , - - ., ,

v,, = E d or

Vl 2 E=-. IP

(5-10)

The total current is

I = J J . ~ ~ = J s _ a ,* . ,

or t ,. , - , .k+rs I .

7 ,: +,

I - < , . , , 1

J = ~ . . (5-1 1)

Using Eqs. (5-10) and (5-11) in Eq. (5-8), we obtain

which is the same as Eq. (5-9). From Eq. (5-12) we have the formula for the resistance of a straight piece of homogeneous material of a uniform cross section for steady current (DC).

We could have started with Eq. (5-9) as the experimental o h m s law and applied it to a homogeneous conductor of length C and uniform cross-section S. Using the formula in Eq. (5-13), we could derive the point relationship in Eq. (5-8).

Example 5-1 Determine the DC resistance of 1 (km) of wire having a I-(mm) radius (a) if the wire is made of.copper, and (b) if the wire is made of aluminum.

Solution: Since we are dealing with conductors of a uniform cross section, Eq. (5 -13) applies.

a) For copper wire, a,, = 5.80 x lo7 (S/m):

G = lo3 (m), S = 7 ~ ( l O - ~ ) ~ = lo-% (m2). We have

. , ., ,~";;; R , = - = lo3 G - - . = 5.49 (n).

a,S 5.80 x lo7 x lo-%

- ' We will discuss the significance of V,, and E more in dctail in Section 5-3.

(5-li)

/-- -12)

2 rem~unce for steady

J (5-13)

d applied it Using the

mm) radius

I : II

b) For a l u k n u h i l re , ual = 3.54 x .lo7 mi: 1 - *

. .

1 I

The conductahcej{G, or the reciprocal of ri%istance, is useful in combining resistances in parallel :

1 S + ' (9. : , G = - = b . -

% I R e (5-14) *

d ft From circuit theory %e know the following: ',

a) When resistakeb R , and R, are connectod in series (same current), the total resistance R is

I (5 - 15)

b) When resistances R , and R, arc connected in parallel (same voltage), we have

5-3 ELECTROMOTIVE FORCE AND KIRCHHOFF'S VOLTAGE LAW .

In Section 3-2 we &inted out that static electric field is conservative and that the scalar line integral a! static electric i?+o-Gty around any closed path 1s zero; that IS,

$ ~ - d t ' = 0 . (5-17)

For an ohmic material J = oE, Eq. (5-17) becomes

Equation (5-18) t1119hs that a steady current cannot be maintained in the same direction in a closed circuit .by an electrostaticjield. A steady current in a circuit is the result of the motion of c ~ a r g e carriers, which, in their paths, collide with atoms and dissipate energy in the circuit4 This energy must cdme from a nonconservative field, since a charge carrier completing a closed circuit in conservative field neither gains nor

178 STEADY ELECTRIC CURRENTS I 5 . ,

1 + + 2 - - ++02 - - - -

+ - + & - + - +E'-

. Fig. 5-3 Electric fields inside an Electric battery electric battery.

loses energy. The source of the nonconservative field may be electric batteries (conversion of chemical energy to ,electric energy), electric generators (conversion of mechanical energy to electric energy), thermocouples (conversion of thermal energy to electric energy), photovoltaic cells (conversion of light energy to elect'ric energy), or other devices. These electrical energy sources, when connected in an electric circuit, provide a driving force for the charge c~rriers. TllisJorcc mmifcsts itself as . an equivalent irnpressed electric field irtterisity E;.

Consider an electric battery with electrodes 1 and 2, shown schematically in Fig. 5-3. Chemical action creates a cumulation of positive and negative charges at electrodes 1 and 2 respectively. These charges give rise to an electrostatic field intensity E both outside and inside the battery. Inside the battery, E must be equal in magnitude and opposite in direction to the nonconservative E, produced by chemical action. since no current flows in the open-circuited battery and the net force acting on the cli;~rgc carriers must vanish. Thc line inlcgrnl of thc impressed ficld intensity E, from the negative to the positive electrode (from electrode 2 to electrodc 1 in Fig. 5-3) inside the battery is customarily called the electromotive forcet (emf) of the battery. The SI unit for emf is volt, and an emf is not a force in newtons. Denoted by V , the electromotive force is a measure of the strength of the nonconservative source. We have

Inside the source

The conservative electrostatic field intensity E satisfies Eq. (5-17).

Outside Inside the source the source

' Also called electromotance.

- :rics (con- w u o n of la1 energy ic energy), in electric ts itself as

n a t d in char- at IC field in- :e equal in ; chemical me actlng 1 ldtensity [rode 1 in ' (emf) of ,. Denoted riservative

(5-19)

-0)

combining Eqs. (5-119) and (5-20), we have .

. , " . ' "

, = J : E ~ (5-21)

Outs~de ' the source

or ir )I

" = v,,= v, - v,. (5-22)

In Eqs. (5-21) and @-22) we have expressed the emf of the source as a line integral of the conservative & and interpreted it as a uqltage rise. In spite of the nonconservative nature of E,, the emf can be expressed as a potential difference between the positive and negative terml?als. This was what we did in arriving at Eq. (5-10).

Whcn a resisto? in tile form of Fig. 5-2 i.p conncctcd bctwecn tcrm~nals 1 '~nd 2 . ol l l x battery, com~lcling the circu~l. lie lotul c l ca r~c field intcnbity (c1ectroat:rtic E caused by charge cumulation, as well as impressed Ei caused by chemical sctlon) must be used in the point form of Ohm's law, We have, instead of Eq. (5-8),

where E, exists inside the battery only, while has a nonzero value both inside and . outside the source. From Eq. (5-23), we obtain

The scalar line integkal of Eq. (5 -24) around the closed circuit yields, in view of Eqs. (5-17) and (5-19),

r - = $(E + E,) . dP = - J . LIP. II Equation (5-25) should be compared to E q (5-18), which holds when there IS no source of nonconse~vative field. If the resistor has a conductivity o, length /, and uniform cross-section S, J = I / S and the right side of Eq. (5-25) becomes RI. We havef

= X I . (5-26)

If there are more than x c soarc; illc!c~.;;,,.~otive force and more than one resistor (including the interdal resistances 01 the sources) in the closed path, we generalize Eq. (5-26) to

-

' We assume the battery to have a degligible 1ntern:l resistance; otherwlu is elfcvt must be lncluded in Eq. (5-26). An idcul witaye solrm in one whosc t c h i n a l Voltage is equal to ltr c n ~ l and is , n d c p c n d c ~ of the current flowing through it. This Implies that an ideal voltage source has a zero internal resistance.

L ' I

L 4 . , I : Equation (5-27) is an expression of Kirchhoff s voltuge l& It states that around a I, I

closed path in an electric circuit the algebraic sum of the emf's (volhige rises) is equal : , . to the algebraic sum of the voltage drops across the resistances. It applies to any closed ' path in a network. The direction of tracing the path can be arbitrarily assigned. and the currents in the different resistances need not be the same. Kirchhoff's voltage law is the basis for loop analysis in circuit theory. I

i i

I . I t

5-4 EQUATION OF CONTINUITY AND . I I

1 KIRCHHOFF'S CURRENT LAW \ < .

L

The priwiplc ofmnrrruotion of cbergr is one of the fundamental postulates of physics.

Electric charges may not be created or destroyed; all charges either at rest or in : motion must be accounted for at dl times. Consider an arbitrary volume V bounded by surface S. A net charge Q exists within this region. If a net current I flows across i Illc surfhce o i r l oS Illis region, the cll;~rge in 1I1e volume inus( ( / ~ ~ c ~ . ~ ~ r a ;I( ;I r;~le tI1;11 t

I

equals the current. Conversely, iSa net currunt llows across the surhce illto the region, - the charge in the volume must i~~creusr at a rate equal to tht-current. The current leaving the region is the total outward flux of the current density vector through the surface S. We have

Divergence theorem, Eq. (2-107), may be invoked to convert the surface integral of , J to the volume integral of V . J. We obtain, for a stationary volume,

ap S V v . ~ d v = - S v z d u . (5 -29) 7

In moving the time derivative of p inside the volume integral, it is necessary to use : partial dinerentiation because p may be a function of time as well as of space co- I

ordinates. Since Eq. (5-29) must hold regardless of the choice of V, the integrands ; must be equal. Thus, we have. I

i h S!

t u (5-30) t - 1

i -7

# - This point relationship derived from the principle of mnxrvation of charge is called the equation of continuity. : I -.. T

For steady currents, charge density does not vary with time, ap/& = 0. Equation , C( (5-30) becomes I . 1(

V . J = O . (5 -3 1) " . -

Thus, steady electric currents are divergenceless or solenoidal. Equation (5-31) 1 - U i is a point relationship and holds also at points where p = 0 (no flow source). It means :

t in

; A 4 1

I

, , < . < ' I

urourid u . , J ~ s e y t t u l , ,

rtly closed , ,

jned, and r

s voltage I

fphysics. est or in . bounded h s across rate that ic rcgion, : current 0 ug'P,2

i

t 5 -28)

ilegral of

(5 -29)

ry to use pace co- tegrands

(5-30)

is c 6 d

:qui

(5-31)

n (5-31) It means

5-4 I EBUA~ON:?F CONTINUITY A/I KIRCHHOFF'S CURRENT LAW 181 d !

i . 4

that the field lines df stryimlines of steady cdrrents close upon themselves, unlike tho& of electrosiatit field intensity that orig!hate and end on charges. Over any enclosed surface, kq. (5-31) leads to the fo l lo~ing integral form:

I $ J - d s - 0 , I (5-32) '

which can be writted as

(5-33)

I 1 . ,

Equation (5-33) is a9exb;kssion of ~ i r c / & f l ' s currrnr iuw. It states that rile o/~ehr?iic sum o/ull the out ($a junction i l l on electric circuit is zero.' KirchholT's

for node analysis in circuit theory. In Section 3-6 ~e stilted that charges introduced in the interior of a coiiductor

will move to the c o d uctor surface and redistribute themselves in such a way as to make p = 0 and E ='! inside h d e r equilibrium conditions. We arc now in a position to prove this statem&nt and to calculate the time it takes to reach a n equilibrium. Combining Ohm's I&, Eq. (5-8) , with the equation of continuity and assuming a constant n, we have

In a simple medium, V E'= p i t and Eq. (5-343 becomes

The solution of Eq. (3-34) is

where po is the initial charge density at r = 0. Both p and po can be functions of the space coordinates, add Eq. (5-36) says that the charge density at a given location will decrease with time cxponentially. An initial charge density po will decay to l/e or 36.8% of its vdue in a time equal to

1- .-. The time constant T is cdled the re/avo: 3,. : i r ~ -or a good conductor such as copper-a = 5.80 x 10' (S/m), r 2 so = 8.85 x lo- ' ' (F/m)- r equals 1.52 x 10-19(s), a very short time indeed. The transient time is so brief that for all practical

' This includes the currcnta of current generators at the junction. if any. An ideal a n e n t genemior s one whose current is independent of its terminal voltage. This implies that an ideal current source h n an Infinite internal resistance.

I purposes p can be considered zero in the interior of a conductor-see Eq. (3-64) in Section 3-6. The relaxation time for a good insulator is not infinite, but can be hours or days.

, 5-5 POWER DISSIPATION AND JOULE'S LAW

In section 5-1 we indicated that under the influence of an electric field, conduction electrons in a conductor undergo a drift motion macroscopically. Microscopically these electrons collide with atoms on lattice sites. Energy is thus transmitted from the electric field to the atoms in thermal vibration. The work An* done by it11 clectric field E in moving a charge q a distance A( is qE . ( A t ) . which co~.responds LO ;L power

where u is the drift velocity. The tot;~l power delivered to all the charge carriers in a volume dv is

which, by virtue of Eq. (5-7), is

Thus the point function E . . I is a power dm,siry rtndcr rtcztdy-currcnt conditions. For a given volume V, the total electric power converted into heat is

This is known as Joule's law. (Note that the SI unit for P is watt, not joule, which is the unit for energy or work.) Equation (5-39) is the corresponding point relationship.

In a conductqr of a constant cross section, dv = ds d t , with d t measured in the direction J. Equation (5-40) can be written as

P = S ~ E ~ / ~ J ~ ~ = V I ,

where I is the current in the conductor. Since V =,RI, we have

(5-41)

Equation (5-41) is, of course, the familiar expression for ohmic power representing the heat dissipated in resistance R per unit time.

1

I )

. 8 . i j3-64) in ' . . 1 t can be ' a

. I

nduction topically . ;ted from' n electric ) 3 power

(5-38)

ricrs in a

'P

(5-39)

ions. or'

. - (5 -40)

\~hich is tionship. ed in the

. - . : I >

I f I

. a - P I 2. (-6 /BOUNDARY CONDI~~ONS F ~ R CURRENT DENSITY 183 ,

~ & - I

I . I A

; 5-6 BOUNDARY COND~TI~NS FOR CURRENT DENSITY * r:

I , L '

A. ' k'

When current .obliqiaely posses an interface4;between two media with difierent conductivities, the cuhent density vector changis both in direction and in magnitude. A set of boundary cohditions can be der~ved fur J in a way similar to that used in Section 3-9 for obt~ining the boundary conditions for D and E. The governing equations for steady currcpt density J in the absence of nonconservative energy sources are I, -

dovernjng Equations for $tea& Current Density

Differential Form lhtegral Form

The divergence equation a the same as Eq. (5-31), and the curl equation is obtalned by combining Ohm'sd lau ( J = oE) with V x E = 0. By applying Eqs. (5-42) a d (5-43) at the interface between two ohmic media with bonductivities o, and oi, we obtain the boundary conditions for the normal Bnd tangential components of J.

Without actually bonstructing a pillbox at the interface as was done in Fig. 3-22. we know from Section 3-9 that the normal rowq>onont of ' l .u diuergence1er.s oectarJe1~1 is continuous. Hence, ham V * J = 0, we have

Equation (<-2vstatei that the ratio o f rhr rangenrial componel~ts of J at tbvo sides of an interface is equal to the ratio of the conductivities.

Example 5-2 Two cbnducting media with contluctivities o, and o, are separated by an interface, as shown in Fig. 5-4. The steady current density in medium 1 at point PI has a magnitude J, and makes an angle a, with the normal. Determine the magnitude and direction of the current density at point P i n medium 2.

184 STEADY ELECTRIC CURRENTS I 5 f?

, . . -

"- ..* ! 1 I

i I

Fig. 5-4 Boundary conditions at interface between two conducting media (Example 5-2). 1

. '

Solution: Using Eqs. (5 -44) and (5-49, we have I

1 ' J, cosa, = J, cosa, (5 -46)

and (5 -47) a 2 4 sin sr, = alJ2 sin a,.

Division of Eq. (5-47) by Eq. (5-46) yields

-- (5 -48) tan a ,

If medium 1 is a much better conductor than medium 2 (a, >> a, or a,/a, 2 0), a, approaches zero and J, emerges almost perpendicular to the interface (normal to the surface of the good conductor). The magnitude of J, is

J 2 = J G J Z = J(J, sin a,)' + (J, cos a,),

= [(: Jl sin c x l r + (Jl cos al l2 I,',

By examining Fig. 3-4, can you tell whether medium 1 or medium 2 is the better conductor? i .

I r7 For a homogeneous conducting medium, the differential form of Eq. (5-43) 1

simplifies to t h v x J = O . (5-50) i s &'; . . / . -

i ': 1 . . . From Section 2-10 we know that a curl-free vector field can be expressed as the r 1 G gradient of a scalar potential field. Let us write

., ,

J = -Vt//. (5-51) t i , 1

1 ; 1

.ce 5 -2).

(5-46)

r.5 -47)

( F 8 )

I + b), g2

lormal to

4

(5 -49)

the bettef

Jq. (5-43) n (5-50)

ied as the

(5-si)

* ,*

Substitution of Eq. (4-51)~hto V . J = 0 yields a Laplace's equation in (I; that is, 1

A problem in steadGburrent flow can therefor& be solved by determining (1 (Aim) from Eq. (5-52), ct to"appropriate boundary conditions and then by finding J from its negative in exactly the same way as a problem in electrostatics is

4 . solved. As a matter 01 fact, ) and electrostatic dotential are simply related: (1 = GV. As indicated in similarity between electrostatic and steady-current

electrolytic tank to map the potential distribution of difficult-to-solve boundary-value problems.'

When a stcady cllrrent flows ilcross thc boundary hetwcen two dillcrent lossy '

dicleclrics (diclcctrics with per~nittivitics e l and r , ilnd finite conductivi[ics n , ;ind n2), tlic L : I I I ~ C I ~ ~ ~ : I I C O I I I ~ O I I C I ~ L 01' tI1c clcct~ic Iicld is C O I I ~ ~ I I L I O U S ;~crobs the intct-hce ;is usual; that is, E,, = El, , which is equivalent to Eq. (5-45). The normal component of the electric field, however, must simultaneously satisfy both Eq. (5-44) and Eq. (3 - 1 13). We require

J l r l - . J Z , , - El, , = qZE2,, (5 5 3 )

Dln - D2n = PA + E ~ E I , ~ - E z E ~ ~ = P s , (5-54)

where the reference &it normal is outward frqm medium 2. Hence, unlsss o,jo, =

c2k,, a surface charbe must exist at the interface. From Eqs. (5-53) and (5-54). we find

Again, if medium 2 is a much better conductor than medium 1 (a , >> a , or ~ , / o , - 0). Eq. (5-55) becomes abproximately

ps = c l E l n = d l , , which is the same as Eq. (3 - 1 14).

Example 5-3 An emf Y is, applied across a parallel-plate capacitor of area S. The space between the conductive plates is filled with two different lossy dielectrics of thicknesses dl and d,, permittivities c l and c,, and conductivities o l and G , respectively. Deteriiiine (a) the current density between the plates, (b) the electric field intensities in both dielectrics, and (c) the surface Charge densities on the plates and at the interface.

' See, for instance, E. Webet, Elecmrqagnefic Fields, Vol. I : Mapping of Fields, pp. 187-193. John Wilcy and Sons, 1950.

, i I .

t . , 1 ;

A:;.c

, . I -. *\'I; :. . "$..'.I' , \ .

. . , / r,

1 1 ' - Y

I ' 5 -

j , - ,[

Fig. 5-5 Parallel-plate I ; a

capacitor with two lossy I r

dielectrics (Example 5-3). I

Solution: Refer to Fig. 5-5.

a) The continuity of the normal component of J assures that the current densities and, therefore, the currents in both media are the same. By Kirchhoff's voltage law we have

and (5-58) o l E l = o z E 2 . (5 -59)

, Equation (5-59) comes from J , = J , . Solving Eqs. (5-58) and (5-59), ive obtain

and

f ' 6 - r --

Ps2 = - c2E2 = - E2019'- ( c / m 2 ) . (5 -63)

62dl + ~ l d 2 I " i'!

<-

f *. ,"*. The negative sign in Eq. (5-63) comes about because E2 and the outward normal I , * <

at the lower plate are in opposite directions. F I

voltage

(5-58)

(5 -59)

e obtain

(5-621

' normal

5-7, / RESISTANCE CALCULATIONS 187

\ '

I

J

. ~ ~ u a t i o d { i - 5 $ ) can be used to find the ikface charge density at the interface of the dielectria $e have

a .

! I ' c 7

From these resultl, we see that p,, # - p ,,,,, but that p,, + p,, + p,, = 0 .

In Example 5-3 wb encounter a situation w i r e both static charges and a steady current exist. As we dball see in Chapter 6, a steady current gives rise to a steady magnetic field. We h d e , thcn, both a static electric field and a steady magnetic field. . They constitute an efdctromugnetostuticfield. The electrlc and magnetlc fields of an electromagnetostatic;fjeld are coupled through the constitutive relation J = aE of the conducting m e d i ~ h .

In Section 3-10 we diicussed the procedure for finding the capacitance between t i o conductors separated by a dielectric medium. ~ h e s e conductors may be of arbitrary shapes, as was shown in Fig. 3-25, which is reproduced here as Fig. 5-6. In terms of electric field quarltitieb, the basic formula for capacitance can be written as

$s D ds $s E E . ds .I c= -= - - I.

(5-65) V - J , E . ~ < -J,E.&' * A

whcrc the surhcc i d ral in thc numerator is carried out over a aurhce enclosing the positive conductor, an the line integral in the denominator is from the negatlve (lower t potential) conductor tb the positive (higher potential) conductor (see Eq. 5-21).

41 I 1 d0 fig. 5-6 Two cnndiictors in r bssy v12 dielectric medium.

When the dielectric medium is lossy (having a small but nonzero a current will flow from the positive to the negative conductor and a current-density field will be established in the medium. Ohm's law, J = cE, ensures that the streamlines for J and E will be the same in an isotropic medium. The resistance between the conductors is

where the line and surface integrals are taken over the same L and S as those in Eq. (5-65). Comparison of Eqs. (5-65) and (5-66) shows the following interesting relationship:

-1 Equation (5-67) holds if r and u of the medium have the same space dependence or if the medium is homogeneous (independent of space coordinates). In these cases, if the capacitance between two conductors is known, the resistance (or conductance) can be obtained directly from the €/a ratio without recomputation.

Example 5-4 Find the leakage resistance per unit length (a) between the inner and outer conductors o fa coaxial cable that has an inner conductor of radius a, an outer conductor of inner radius b, and a medium with collductivity 0; and (b) o f a parallel- wire transmission line consisting of wires of radius a separated by a distance D in a medium with conductivity a.

Solution

a) The capacitance per unit length of a coaxial cable has been obtained from Eq. (3-126) in Example 3-16.

Hence the leakage resistance per unit length is, from Eq. (5-67),

I ( ) = n ) * CI (Q/m). - (5-68)

' The conductance per unit length is GI = l/R,. '

T

T'

It A1gtl.

Agtk

. Ir a. -!@! curre1 comp be cor const:

h frinplr TI

betwe~

1. Ci 2. 4. 3. Fi

ge e 'I E

4. Fl P

i-P ? - Fl

I t is ir; contlu hold. circurr

he stream- ttween the

(5 -66)

lose in Eq. interesting

.s/' 1denG or if :ase. he ncc) can be

: inner and a, an outer

T a parallel- m e D inla

d from Eq.

n.

I

, . ', . i;

' b) For the parallel-~ire transmission link, Eq;*(4-47) in ~ x a m ~ l e 4-4 gives the . capacitance per Lihit length. '

I , I

' '# C.

nE c; = (F/m). cosh- l (&-

b

Therefore, the ledcage resistance per unit length is, without further ado,

'I'hc cunductuncc per un i t Icnytli is C;, -: 1/1( ,. It must be empHasized here that the resistance betweer1 the conductors for a

length L of the coaxidl cable is R J t , not LR,; similarly, the leakage resistance of a length L of the parallel-wire transmission line is R'JL, not CR',. Do you know ~vhy.?

In certain situations, lcctrostatic and steady-current problems are not ex:<cily ;m~logous, even wheh the gcotnctrical conljgurations are the same. This is becatise current tlow can be tonlin?d strictly within a conductor (which ius a wr.!, l o r q ~ ri

compared to that of'the surrounding medium), whereas electric flux usually cannot be contained within EL dielectric slab of finite dimensions. The range of the dielectric constant of available materials is very limited (see Appendix B-31, and the flux- fringing around conductor edges makes the computation of capacitance less accurate.

The procedure for computing the resistance of a piece of conducting material between specified eqbipotential surfaces (or terminals) is as follows:

1. Choose an apprdpriate coordmate system for the given geometry.

2. Assume a potential diRerence V, between conductor terminals.

3. Find elects~c ficld ultcnsity E w1t11111 I ~ C conductor. (lf thc m,iter~'ll 1s homogeneous, having a cousratlt conductivity, the gcncral method 1s to mlve Lapl'lcc's equation V2V = 0 for V in the chosen coordinate system, and then obtam E = -VV.)

4. Find total currefit

where S is the cross-sectional area over whir b : I flows. n )

5. Find resistance R by taking the ratio V , , / I . .

It is important to note that if the conducting material is inhomogeneous and if the conductivity is a function of space coordinates, Laplace's equation for V does not hold. Can you expfhin why and indicate how E can be determined under these circumstances?

L

, J , < b #,.

;r-, t i . ' - . . a . ., . When the given geometry is such that J can be determined easily from a total ." ,+ .: r current I, we may start the solution by assuming an I. From I, J and E = J b are . 4;

found. Then the potential difference Vo is determined from the relation , ,

where the integration is from the low-potential terminal to the high-potential terminal. The resistance R = Vo/I is independent of the assumed I, which will be canceled in the process.

I

- Example 5-5 A conducting material of uniform thickness h and conductivity u , h has the shape of a quarter of a flat circular washer, with inner radius a and outer radius b, as shown in Fig. 5-7. Determine the resistance between the end faces.

I

i Solution: Obviously the appropriate coordinate system to use for th i~ '~roblem is f the cylindrical coordinate system. Following the foregoing procedure, we first assume a potential difference V, between the end faces, say V = 0. on the end face at y = 0, and V = Vo on the end face at x = 0. We are to solve Lapluce'Sequ:lfion in V subject to the following boundary conditions:

Since potential V is a function of 4 only, Laplace's equation in cylindrical coordinates simplifies to

d2V -- d4' - O' (5 -7 1)

The general solution of Eq. (5-71) is

V = c14 + C2,

which, upon using the boundary conditions in Eqs. (5-70a) and (5-70b), becomes

.-" ,W<! . . *-* $

. - . Fig. 5-7 A quarter of a flat circular 0 X washer (Example 5-5).

terminal. lceled in

! .

I I (.

, < l

ctivity a i ' ~d outer '

:es.

~blem is assume

i t y = 0, ' aubject

P

15--

(5-70b)

.dinates

( 5 -7 13

:comes

(5,-72)

n I

\ .

i I : RFVlEW QUESTIONS 191 . . . .

The total current, I b n be found by integrating J over the 4 = n/2 surface at which ds = - a6h dr. We have

J

Therefore,

Note that, for this problem, it is not convenient to begin by assuming a total current I because it is not obvious how J varies with r for a given I without' J, E and Vo cannot be ddtermincd.

REVIEW QUESTIONS I ?

R.5-I Explam the dikerewe between c o ~ d u c t ~ o n and convect~on currents.

R.5-2 Explain the dbera:ion of on electrolytic tank. In what ways do electrolytic currents differ from conduction and convection currents?

R.5-3 What is the point f x m for Ohm's law?

R.5-4 Define conductivity. What is ~ t s SI unit?

R.5-5 Why does the Rsisiance fornlula in Eq. (5-13) require that the material be homogeneous and straight and that it hav: a uniform cross section?

R.5-6 Prove Eqs. (5-15) and (5-16b).

R.5-7 Define electroHtotive Jorce in words.

R.5-8 What is the dikrencc between impressed and electrostatic field intensities?

R5-9 ~t'a*~irchhoff's vdtage law in words.

R5-10 What are the charactiristics of an ideal voltage source?

115-11 Can the currehts in different branches (resistors) of a closed loop in an electric network flow in opposite directions? Explain.

R.5-12 .What is the pHpica1 significance of the equation of continuity?

192 STEADY FLECTRIC CURRENTS I 5

R5-13 State Kirchhoff's current law in words.

R5-14 What are the characteristics of an ideal current source?

R5-15 Define relaxition time.

R.5-16 In what ways should Eq. (5-34) be modified when a is a function of space coordinates?

R.5-17 State Joule's law. Express the power dissipated in a volume

a) in terms of E and a, b) in terms of J and a.

R5-18 Does the relation V x J = 0 hold in a medium whose conductivity is not constant? Explain. I

R.5-19 What arc the boundary condilions of the normal and tangential components of steady i

current at the interface of two media ~ 5 t h different conductivities?

R.5-20 What is the basis of using an electrolytic tank to map the potential distribution of elec- i [rcwttic hound;~ry-v:dw problen~s? '

'-\

H.5-21 WIi;tt is tllc rclation bclwcen the rcsis[;~ncc and the capacitace fortil~d by tivu con- --. ductors immersed in a lossy dielectric medium that has permittivity E and conductivity n?

R5-22 Under \ ~ h i ~ t situations will the relati011 between Rand C in R.5-21 be only approrimatcly correct? Give a specific example.

PROBLEMS

P S I Starting with Ohm's law as expressed in E q (5-12) applied to a resistor of length t . conductivity a, and uniform cross-section S, verify the point form of Ohm's law represented by i

Eq. (5-8).

P.5-2 A long, round wire of radius a and conductivity 0 is coated with a material of conductivity 0. la.

a) What must be the thickness of the coating so that the resistance per unit length of the uncoated wire is reduced bv SO%?

b) Assuming a total currenf I in the coated wire, find J and E in both the core and the coating material.

Fig. 5-8 A network problem (Problem PJ-3).

1

t constant?

J tts ut' steady

oy two coh- it? a P

of length C, presented by

length bf the

core and the

r PROBLEMS 193

i , I

P.5-3 Find the current h d the heat dissipated in each of the five resistors In the network shown " A . - in Fig. 5-8 if

5 I

4 J

R l = 4 (a), ; R 2 = 20 (Q), R, = 30 (Q), R, = 8 (Q), R 5 = 10 (a) ,

and if the source is ad idebl D C voltage generator of 0.7. (V) with its positive polarity at terminal 1. What is the total resistdnce seen by the source'at terrhinal pair 1-2?

P.5-4 Solve problem PS-3, assumlng the source IS hd ideal current generator that supphes a direct current of 0.7 (A) out of terminal 1. q

7 P.5-5 Lightning strike&,a lossy dielectric sphere-€ 1.2 E,, a = 10 (S/m)-of radius 0.1 (m) at time t = 0, depositing pnifo;mly in the sphere a totkl charge 1 (mC). Determine, for all r,

a) thc electric ficld htensiiy both inside and outsihc the sphere, 1)) the current dcnslty in tlie sphere.

P.5-6 Ilcfer to I'roblcm 1'5-5. a) Calculate the t h e it takes for the charge denshy in the sphere to diminish to 1 7 , of its

initial value. 1 1

b) Calculate the change in the electrostatic energy stored in the sphere as the charge density diminishes from the nitial value to 1 % of its value. What happens to this energy'? .

c) Determine the e1ect:osratic energy stored in the space outside the sphere. Does this energy change with time?

P.5-7 A D C voltage af 6 ( V ) applied to the ends of 1 (km) ? f a conducting wire of 0.5 (mm) radius results in a curred of 116 (A), Find

a) the conductivity of the wire, . b) the electric field hteilsity in the wire, c) the power dissiphted in the wire.

a) Draw the eq~ivhlent circuit of the two-layer, barallel-plate capacitor with lossy dielectrics, and identify the magnitude of each component.

b) Determine the Ijbwer dissipated in the capacitdr.

P.5-9 An emf Y' is appliec across a cylindrical capacitor of length L. The radii of the inner and outer conductors are a a i d 11 ruspectively. The space between thc conductors is fillcd with two different lossy dielectric.; having, respectively, permittivity E , and conductivity a , in the region a < r < c, and permittivity € 2 and conductivity a, in the region c < r < h. Determine

l

a) the current density in each region, b) the surface charge densities on the inner and outer conductors and at the interface between

the two dielectrics. ---. P5-10 Rekr to the flat quaricr-circular washer In Example 5-5 and Fig. 5-7. Find the resistance

between the curved sldes.

P.5-11 Determine the resi~tance between concenttic spherical surfaces of radu R l and R2 ( R , < R,), assuming that a material of conductivity a = a,(l + k/R) fills the space between them. (Note: Laplace's equation for V does not apply here.)

194 - STEADY ELECTRIC CURRENTS 1 5

, t . ,

1 /,.& > . P5-12 A homogeneous material of uniform conductivity t; is'shaped like a truncated conical block and defined in spherical coordinates by

+ R , I R S R , and 0 1 0 ~ 8 , .

Determine the resistance between the R = R , and R = R, surfaces.

P.5-13 Redo problem P.5-12, asguming that the truncated conical block is composed of an inhomogeneous material with a nonuniform conductivity a(R) = aoRl/R, where R , 5 R I R,.

I P5-14 Two conducting spheres of radii b, and b, that have a very high conductivity are immersed

i A %

in a poorly conducting medium (for example, they are buried very deep in the ground) of conductivity a and permittivity e. The distance, d, between the spheres is very large compared with the radii. Determine the resistance between the conducting spheres. Hint: Find the capacitance between the spheres by following the procedure in Section 3-10 and using Eq. (5-67).

/ ' I

P.5-15 Justlfy the statement that the steady-current problem associated with a conductor t

buried in a poorly conducting medium near a plane boundary with air, as shown in Fig. 5-9(a), can be replaced by that of the conductor and ~ t s image, both immersed in the poorli conducting ! medium as shown in Fig. 5-9(b).

Q Boundary removed o = o ----------- I

u d Fig. 5-9 Steady i 0 current problem with a

plane boundary *-a *A.e .. . -. .-.. .a.4 ,..... . . I I.U (Problem P.5-15). I

! (a) Conductor in a poorly (b) Image conductor in conducting.

conducting medium near medium replacing the a plane boundary. plane boundary.

P5-16 A ground connection is made by burying a hemispherical conductor of radius 25 (mm) in the earth with its base up, as shown in Fig. 5-10. Assuming the earth conductivity to be

S/m, find the resistance of the conductor to far-away points in the ground.

Fig. 5-10 Hemispherical conductor in ground (Problem P.5-16).

P.5-17 Assume a rectangular conducting sheet of conductivity u, width a, and height b. A - , potential difference V, is applied to the side edges, as shown in Fig. 5-11. Find .-..

a) the potential distribution b) the current density everywhere within. the sheet. Hint: Solve Laplace's equation in

Cartesian coordinates subject to appropriate boundary conditions. 1 i

sed of an i R 5 R 1 .

~mmerscd d) of con- lared with '

I .

spacitance 1.

conductor -ig. 5-9(4, zonducting

/-'

:ad y em with a lry -15).

%us 25 (mm) ~ctivity to be

f-

d heirL+ b. A $*.

's equation in

j r'

v = 0 -- = an

I I Fig. 5-11 A conducting sheet 1- a ---7 (Problem P.5-17).

- ,

P.5-18 A uniform curreht density J = a,Jo flows in a k r y large block of homogeneous material of conductivity a. A hole of radius b is drilled in the material. Assuming no var~ation in the =-direction: find the nevJ current density J' in the cdllducting material. H~nt: Solve Laplace's equation in cylindrical &ordinates and note that V approaches -(Jor/a)cos$ as r -+ a.

6 / Static M ngnetic Fields

6-1 INTRODUCTION f

In Chapter 3 we dealt with static electric fields caused by electric charges at rest. We saw that electric field intensity E is the only fundamental vector fikld quantity required for the study of electrostatics in free space. In a material medium, it is convenient to define a second vector field quantity, the electricJlx density D, to account for the effect of polarization. The following two equations'form the basis of the electrostntic model :

V .U=, , (6- 1)

The electrical property of the medium determines the relation between D and E. If the medium is linear and isotropic, we have the simple cortstit~itiue rclurion D = EE.

' When a small test charge q is placed in an electric field E, it experiences an

electric force F,, which is a function of the position of q. We have

When the test charge is in motion in a magnetic field (to be defined presently), experiments show that it experiences another force, F,,,, which has the following characteristics: (1) The magnitude 'of I?, is proportional to q ; (2) the dircction of I;, at any point is at right angles to the velocity vector of the test charge as well as to a fixed direction at that point; and (3) the magnitude of F, is also proportional to the component of the velocity at right angles to this fixed direction. The force F, is a magnetic force; it cannot be expressed in terms of E or D. The characteristics of F, can be described by defining a new vector field quantity, the magnetic Jux density B, that specifies both the fixed direction and the constant of proportionality. In SI units, the magnetic force can be expressed as

?::J$. . ,

(6-4)

6 -2 MAGE

r s at rest. i quantity . i t is con- o account .sis @he

(6-1)

\ .2)

) and E. If )n D = EE. :riences an

tly), experi- ,

: character- I' F, at any s to a fixed to the com- 4 a magnetic : F, can be 3sityFthat ;I ur. , the

. 1 I: I

where u (m/sJ is thk vdocity vector, and B is mkasured in webers per square meter (Wb/m2) or t e d d 0,' The total eleetrohag$etie force on a charge q is, then, F = F e + F m ; t h a t i s ,;

(6-5)

which is called Lotdm's jhrce equation. Its kilidity has been unquestionably cstabiishcd by experinlcnts. We may tonbider Fe/q for a small q as the definition for electric field intensity ,k (as we did in Eq. 3-2) and Fm/q = u x B as the defining relation for magneiic t ux density B. Alternative y, we may consider Lorentz's force I equation as a [undadental postulate of our elktromagnetic model; it cannot be derived from ofher postulates,

We begin the stlldy of static magnetlc fields in free space by two postulates specifying the divergedce and the curl of B. From the solenoidal character of B. a vector magnetic potential is defined, which is shown to obey a vector Poisson's equation. Neit we deiive the Biot-Savart law, *hich can be used to determine the magnetic field of a curf'eni-carrying circuit. The postulated curl relat~on leads d~rectly to Ampire's circuital law which is particularly useful when symmetry exists.

The macroscopic bffe-t of magnetic materials in a magnetic field can be studled by defining a magnetihation vector. Here we introduce a fourth vector field quantity, the magnetic field intensity H. From the relation between B and H, we define the permeability of the nl$terial, following which .we discuss magnetic clrcults and the microscopic behaviorlof magnetic materials. We then examine the boundary conditions of B and H at the ~pterface of two different magnetic media; self- and mutual inductances; and magfieti; energy, forces, and torques.

,

6-2 FUNDAMENTAL POST~JLCTES OF MAGNETOSTATICS IN FRE$ SPACE

To study magnetostatjcs (;;teady magnetic fields) in free space, we need only consider the magnetic flux dendity iector, B. The two fundamental postulates that specify the divergence and the curl of B in free space are

-

' One weber per square meter or one I& y u a l s la4 y u s s in CCS units. The earth rnagnetlc field IS about 4 gauss or 0.5 x lo-' T. ( t , weber is the same as a volt-second.)

198 STATIC MAGNETIC FIELDS 1 6

In Eq. (6-7), po is the permeability of free space / (see Eq. 1-9),'and J is the current density. Since the divergence of the curl of any vector field is zero (see Eq. 2- l37), we obtain from Eq. (6-7)

which is consistent with Eq. (5-31) for steady currents. Comparison of Eq. (6-6) with the analogous equation for electrostatics in free

space, V . E = p/eO (Eq. 3-4), leads us to conclude that there is no magnetic analogue for electric charge density p. Taking the volume integral of Eq. (6-6) and applying divergence theorem, we have .

where the surface integral is carried out over the bounding surface of an arbitrary volume. Comparing Eq. (6-8) with Eq. (3-7). we again de$the existence of isolated magnetic charges. There are no magnetic Jow sources, and the magnetic J u x lines always close upon themeloes. Equation (6-5) is also referred to as an expression for the law of conservation of rnagnetic Jux, because it states that the total outward magnetic flux through any closed surface is zero.

The traditional designation of north and south poles in a permanent bar magnet does not imply that an isolated positive magnetic charge exists at the north pole and a corresponding amount of isolated negative magnetic charge exists at the south pole. Consider the bar magnet with north and south poles in Fig. 6- l(a). If this magnet is cut into two segments, new south and north poles appear and we have two shorter magnets as in Fig. 6-l(b). If each of the two shorter magnets is cut again into two segments, we have four magnets, each with a north pole and a south pole as in Fig. 6-1(E). This process could be continued until the magnets are of atomic dimensions; but each infinitesimally small magnet would still have a north pole and a south pole.

Successive division

ics in free1 analogue ~ P P ~ Y ~ Q

-(6 -8)

arb~trary > f ~srficd f lux ..,ies ession for

I or .d

ar magnet 1 polc and outh pole. magnct is vo shorter I into two as in Fig.

msnsions; outh pole.

n

flux lines follow er end outside the magnet, and then The designation of north and south

ends of a bar magnet freely and south directions.

in Eq. (6-7) can be obtained by integrating Stokes's theorem. We have

" , . i ( V x B ) - d ~ = ~ , , * f J . d s . . or ! s

B - dP= poi, (6-9)

I

whcrc thc path C for !he line ihtcgral ia thc contour boundlng the iurlicc S, m d I is the total current t S The sense of tracing C and the dl~ection of current flow follow the Equation (6-9) is a form of Ampere', c ~ r c s ~ t n l lniv, which states that the ~irculation of the lnayrietic f lex doisltj! in jrrc space nroiinii oiijl

I closed path is equal to po tunes the rotol current jawing through die u t jace bounded by the path. -4mpkre9J circuital law is very usefbl in determining the maqnrtlc flux density B caused by LI cuirent I when there is a closed path C around the current - such that the magnitdtle cf R is constant over the path.

Thc followihy is a'sunrmary of the two fundamental postulates of magnetostat~cs in free space:

Free Space

Example 6 4 An infinitely long, straight conductor with a circular cross section --.. of radius b carrles a 3teady current I. Determine the magnetic flux density both

inside and outside the contiuctor.

Solution: First we note tliat this is a problem with cylindrical symmetry and that Ampkre's circuital law can beused to advantage. If we align the conductor along the "axis, the magnetic fldx density B will be &directed and will be constant along any

t

circular path around the z-axis. Figure 6-2 shows a cross section of the conductor !

and the two circular paths of integration. C , and C,, inside and outside, respectively, the current-carrying conductor. Note again that the directions of C , and C , and the : direction of I follow the right-hand rule. (When the fingers of the right hand follow the directions of C , and C,, the thumb of the right hand points to the direction of I . )

a) Inside the cortductor: -1.

B l = a g B g l , de=a,,,r,drl,

$c, B , - d f = So2" B,,r1 dg = 2nr lBgi .

The current through the area enclosed by C , is

Therefore, from Ampkre's circuital law, f

liorlI rl 5 b. t Bl = .,B,l = a, j-&. (6-10) 1 C

b) Outside the conductor: , ! c

B2 = a,Bg2, d t = a,r2 d 4

$B2-dP=Znr,B, , .

Path C2 outside the conductor encloses the total current I. Henc?

Examination of Eqs. (6-10) and (6-1 1) reveals that the magnitude of B increases i 1 -

linearly with r , from 0 until r , = h, after which it decreases inversely with r,.

Example 6-2 Determine the magnetic flux density inside a closely wound toroidal coil wilh an air core having N tur t~s a n d cilrryillg ;I c~trrcI11 I. Tllc 101.0itl I I ~ I S a lncilrl

radius b and the radius of each turn is u. i I

1

4 . 1 .

I .

conductor :spectively, 4 1 C2 dnd the dnd follow a ion of I.) n 1 1

(6- 10)

i 1

I i

r'.

(6-1 1)

3 Increases -I r 2 .

I d toroidal as a mean

i I

1 i 1 1

6-2 1 FUI: AMENTAL POSTULATES OF'MAG~~ETOSTATICS IN

: < FREE SPACE

Fig. 6-3 Current-carrying toroidal coil (Example 6-2). I

Solution: Figure 6-3 drpicts the geometry of this problem. Cylindrical symmetry ensures that B has only a +-component and is cbnstant along any circular path about the axis of the toroid; We constructa circular contour C with radius r as shown. For (b - a) < r < b + a, kq. (6-9) leads diredtly to

$ B - d~ = 2nr~,= ~ , N I ,

where we have assumed that the toroid has an air core with permeability p,. Therefore,

I ' o N I , (b - o) < r < (b + a). - B = asB4 = a4 - (6-12)

2zr

It is apparent that = 0 for r < (b - a) and r > (b + a), since the net total current enclosed by a contour constructed in these two regions is zero.

Example 6-3 Determine the magnetic flux density inside an infinitely long solenoid with an air core having n closely wound turns per unit length and carrying a current I.

Solution: This problem can be s'olved in two ways.

a) As n dire&application ofAmpere'scircuita1 law. It is clear that there is no magnetic field outside of the holenoid. To determine the B-field inside we construct a rectangular contour C of length L that is pdttially inside and partially outside

, the solenoid. By reason of symmetry, the B-field inside must be parallel to the axis. Applying Ampkre's circuital law,,we have

202 STATIC MAGNETIC FIELDS I 6

la Fig. 6-4 Current-carrying long solenoid (Example 6-3).

The direction of B goes from right to left, conforming to the right-hand rule with respect to the direction of tjle current I in the solenoid, as indicated in Fig. 6-4.

b) A s a speciul cux of toroid. Thc straight solenoid may be regarded .as a special case of the toroidal coil in Example 6-2 with an infinite radius (b + cc). In such a case, the dimensions of the cross section of the core are very small compared with b, and the magnetic flux density inside the core isa~proximately constant. We have, from Eq. (6-12).

which is;the same as Eq. (6-13). The &directed B in Fig. 6-2 now goes from right to left, as was shown in Fig. 6-3.

6-3 VECTOR MAGNETIC POTENTIAL

The divergence-free postulate of B in Eq. (6-6), V - B = 0, assures that B is solenoidal. As a consequence, B can be expressed as the curl of another vector field, say A, su'ch that (see Identity 11, Eq. (2-137), in Section 2-10)

The vector field A so defined is called the vector magnetic potential. Its SI unit is weber per meter (Wb/m). Thus, if we can find A of a current distribution, B can be obtained from A by a differential (or curl) operation. This is quite similar to the introduction of the scalar electric potential V for the curl-free E in electrostatics (Section 3-S), and the obtaining of E from the rclation 15 = - V V . Ilowcvcr, the definition of a vector requires the spec ib t ion of both its curl and its divergence. Hence Eq. (6-14) alone is not sufficient to define A; we must still specify its divergence.

How do we choose V A? Before we answer this question, let us take the curl of B in Eq. (6-14) and substitute it in Eq. (6-7). We have

rule with Fig. 6-4.

special 111 such

:mpared : o n s p t .

les from

c~loidal. A. such

unit is can be to the

Jsq- .cr, ~..e rgence. rgel,. curl of

I , 6-3. I V&TOR MAGNETIC POTENTIAL 203

I

Here we digress t6 inhoduce a formula for the curl curl of a vector: .

f v 2 ~ = v ( v . l i ) - V X V X A . (6-16b)

Equation (6-16a)' or (6-16b) can be regarded as the definition of V'A, the Laplacian -, "

of A. For Cartesian &ordinates, it can be rekdily verified by direct substitution (Problem P.6-10) that ,

2 V2-4 = ax VIA, + a , , ' v 2 ~ , + az VIA=. ;.! (6-17)

Thus for Cartesian coordinates, the ~ a ~ l a c i a n of a vector field A 1s another vector field whose componehts are the Laplaclan (the divergence of the gmdlent) of the corresponding components of A. This, howevbr, is not true for other coordinate systems.

We now expand V x V x A in Eq. (6-15) according to Eq. (6-16a) iind obtun

V(V.A) - V% = p o J . (6-13) Wlth the purpose oidmpl~fying Eq. (6-18) ta the greatest extent possible, we choose:

/ V . ~ = O , /

and Eq. (6- 18) becomes

This is a vector IJoissn's quutiotz. In Cartesian Coordinates, Eq. (6-20) is equivalent to three scalar Poisson's zquations:

V'A ,. = - 11, J,, , (6-21b) V2A, = - p o J , . (6-21~)

Each of these three equations is mathematically the same as the Poisson's equation, Eq. (4-6), in electrostatics. In free space, the equation

' Equation (6-Iba) can also be obtained heuristl~llly from the vector triple product formula in E q (2-20) by $onsidering the del operator, V, a vector:

V X (V x ~j = V(V . A) - (V.. V)A = V(V . A) - V2A.

: Equation (6-19) holds 10; static magnetic fields. Modification is necessary for time-varylnp electm- magnetic fields (see Eq. 7-46),


.,-

s , has a particular solution (see Eq. 3-56),

Hence the solution for Eq. (6-21a) is

We can write similar solutions for A, and A:. Combining the three components, we have the solution for Eq. (6-20):

Equation (6-22) enables us to find the vector magnetic potential A from the volume current density J. The magnetic flux density B can then beobtained from V x A by differentiation, in a way similar to that of obtaining the static electric field E from - vv.

Vector potential A relates to the magnetic flux @ through a given area S that is bounded by contour C in a simple way:

@ = B - d s . b (6 -23) The SI unit for magnetic flux is weber (Wb), which is equivalent to tesla-square meter (T.m2). Using Eq. (6-14) and Stokes's theorem, we have

(D = 1 7 ( V x A). d s = $ A . dP (Wb). (6-24)

6-4 BIOT-SAVART'S LAW AND APPLICATIONS

In many applications we are interested in determining the magnetic field due to a current-carrying circuit. For a thin wire with cross-sectional area S, dv' equals S dd ' , and the current flow is entirely along the wire. We have

J dv' = J S dP' = I d t ' , (6-25) and Eq. (6-22) becomes

where a circle has been put on the integral sign because the current I must flow in

ments, we

( 6 - 2 2 )

li. \.olume

- '" "bb' m

j t' is

(6-23)

!d-4 luare &

(6-24)

due to a als S dL',

(Yv)

(6-20)

: flow in

' L' ', 6-4 1 6101-Si VART'S LAW AND APPLICATIONS 205

v .' 4 .', < 4

' i: a closed path,' whicn is designated C'. The mdgnetic flux density is then

, B = V x A = V x - 5 [kt%, %"I

(6-27) I

It is very important 10 note in Eq. (6-27) that the unpiinwf curl operation implies differentiations with respect to the space coordjnates of the field point, and that the integral operation is'hith :iespect to the brimed source eoordmates. The integrand in Eq. (6-27) can be dipanrted into two terms by using the follow~ng identity (see Problem P.2-26): '

' i X ( f G ) = f V x G + ( V f ) x G . (6-38)

Wc liavc, with f = ILH and G = df',

I3 = $ [LV x dP1 + /In: ( ' R 16-29)

Now, slnce the unprlhed .ind primed coordinates are independent. V x d P f equals 0. and the first term on the rizht side oSEq. (0-29) vanishes. The d~stance R IS me~buied from dt" at (x', y', z') to the field point ilt (x, y. :). Thus we have

i = [ (x - .q2 + ( y - yi)' + ( z - _,)?I - 1 2; R

- - - ax(" - x') + a,(y - y') + a,(? - 3')

[(x - 7' + 0' - y')' + ( z - z ' ) ' ]~ '~

R -= 1 R j - -a, 7, R (6-30)

where a, is the unit vector directed j m n the source point to the field point. Substituting Eq. (6-30) in Eq. (6-19), me get

4R ce (6- 3 1)

Equation (6>N).is known as Biot-Souart's law. It is a formula for determining B caused by a current I in a closed path C', and is obtained by taking the curl of A in Eq. (6-26). Sometimes it is convenient to write Eq. (6-31) in two steps.

' w e are now dealing with direct (non-time-varying).currehts that give rise to steady magnetlc fields. Circuits containing time-vdrying sources may send time-varying currents along an open wire and deposit charges at its ends. Antennas are examples.

206 S T A ~ C MAGNETIC FIELDS 1 6

t ,

(6- 32)

with

which is the magnetic flux density due to a current element I dt". An alternative and sometimes more convenient form for Eq. (6-33a) is

Solution: Currents exist only in closed circuits. Hence the wire in the present problem must be a part of a current-carrying loop with several straight sides. Since we do not know the rest of the circuit, Ampere's circuital law cannot be used to advantage. Refer to Fig. 6-5. The current-carrying line segment is aligned with the ;-axis. A typical element on the wire is

dP' = a, dz'.

The cylindrical coordinates of the field point P are (r, 0, 0).

Fig. 6-5 Current-carrying straight wire (Example 6-4).

/ ( , ) I tic' x N = ( 3 ) m i I Comparison of Eq. (6-31) with Eq. (6-9) will reveal that Biot-Savart law is, in

general. more difficult to apply than Ampere's circuital law. However. AmpCre's circuital law is not useful for determining from I in a circuit if a closed path cannot

-. be found over which B has a constant magnitude.

Example 6-4 A direct current I flows in a straight wire of length 2L. Find the magnetic Hux density B at a point located at a distance r from the wire in the bisecting plane: (a) by determining the vector magnetic potential A first, and (b) by applying Biot-Savart's law.

(6-32)

(6-33a)

itive and

(0--3ib)

t i v i \ , in

'-FS CJ. c

IhL .;

:lbeCilllg ipplying

~ o b l e m , : do not vantage. -axis. A

P

I

! 6-4 I B I O T - ~ V A ~ T ' S LAW AND APPLICATIONS 207

r - a) By jinding B frod V x A. Substituting R ,/- into Eq. (6-261, we have

I L dr' i A = ~ Z K J - ~ ,/-

Cylindrical symmetry around the wire assures that BA,/d$ = 0. Thus,

When r << L, Eq. (6-35) reduces to

which is the expression for B at a point located at a distance r from an infinitely long, straight wire carrying current I.

b) By applying Biot-Gauart's law. ~ r o ; Fig. 6-5, we see that the distance vector from the source ekment dz' to the field point P is

R = a,r - a:=' ;+ dt" x R = a= dz' x (arr - aZr1) = a,r dz'.

Substitution in Eci, (6-33b) gives

. which is the same as Eq. (6-35).

Example 6-5 Find the magnetic flux density at the center of a square loop. with side w carrying a direct current I . .

Fig. 6-6 Square loop carrying Y current I (Example 6-5).

a Solution: Assume the loop lies in the xy-plane, as shown in Fig. 6-6. The magnetic flux density at the center of the square loop is equal to fo#,r times that caused by a single side of length w. We have, by setting L = r = w/2 in gq. (6-35).

A

where the direction of B and that of the current in the loop follow the right-hand rule.

Example 6-6 Find the magnetic flux.density at a point -&he axis of a circular loop of radius b that carries a direct current I.

Solution: We apply Biot-Savart's law to the circular loop shown in Fig. 6-7.

Again it is important to remember that R is the vector from the source element dl" to the field point P. We have

Y Fig. 6-7 A circular loop carrying

x currcnt I (Example 6-6) .

nagnetic \cd by a

(6-37)

nd rule. r circular

7 .

.ent de'

. 'i < , . 8-5 1 THE MAGNETIC DIPOLZ 209

- : , ,

i ! ' Because of c$ihdr!cal symmetry, it is easy to see that the a,-component is canceled by the contribution ohthe element located dianletrically opposite to dt", so we need only consider the a,-cbmponent of this cross product.

We write, fford EQ. (6-jl),

6-5 THE MAGNETIC DIPOLE

Wc begin this scction with an cxamplc.

Example 6-7 Find the magnetic flux density at a distant point of a small circular loop of radius b that carries current I.

I

Solution: It is appaIent from the statement of the problem that we are interested in determining B at a point whose distahce, R, from the center of the loop satisfies the relation R >> b; that being the case, we may make certain simplifyin- approximations.

We select the centgt of the loop to be the origin of spherical coordinates. as shown in Fig. 6-8. The s o u t b coordinates are primed. We first find the vector magnetic potential A and then determine B by V x A .

A small circular (Example 6-7).

loop carrying


Equation (6-39) is the same as Eq. (6-26), except for one importart point: R in ! Eq. (6-26) denotes the distance between the source element dt? at P' and the field point P ; but it must be replaced by R , in accordance with the notation in Fig. 6-8. Because of symmetry, the magnetic field is obviously independent of the angle $ of ,

the field point. We pick P(R, 0, n/2) in the yz-planc. Another point of importance is that a,. at de' is not the same as a, at point P

In fact, a, at P, shown in Fig. 6-8 is -a,, and I

dl" = (-a, sin 4' + a, cos 4')b d@. (6 - 40) I

i

For every I dP' there is another symmetrically located differential current element on the other side of the y-axis that will contribute an equal amount to A in the -a, direction, but will cancel the contribution of I de' in the a, direction. Equation (6-39) can be written as

d4 ' 1

or

The law of cosines applied to the traingle OPP' gives

R f = R~ + b2 - 2bR cos +, where R cos + is the projection of R on the radius OP', which is the same as the ,

projection of OP" (OP" = R sin 6) on OP'. Hence,

R f = R2 + b2 - 2hR sin 0 sin 4' and

1 1 b2 2b - = - (1 + -T - - sin 0 sin 4' Rl R R R > - ' I 2

When R2 >> b2, b 2 / ~ ' can be neglected in comparison with 1.

1 b g - (1 + - sin 0 sin 4'

R R

Substitution of Ey. (G-42) in 13q. (6 . 41) yiclcls

b A = a, * Jni2

(1 + - sin 6 sin 4' 2nR R ,

oint: R in d the field 1 Fig. 6-8. angle 4 of

it point P.

(6 - 40)

lement on ! the -a, 1011 (0 .39)

r 1)

TL' ,is the

(6-42)

0

(6-43)

C, I . , + 6-5 / THE MAGNETIC DIPOLE 211

I

The magnetic flux density is B = V x A. Equation (2-127) can be used to find

/f lb2 B=O 4R3

(a, 2 cos 0 + a, sin O), (6-43) , which is our answer.

At this point we recognize the sin~ilarity between Eq. (6-44) and the expression for the electric field intensity in the far field of an electrostatic dipole as given in Eq. (3-49). To exanline the similarity further, we rearrange Eq. (6-43) as

is defined as the rniy(t1eti~ dipole inotncric, whlch is a vector whose ma, onltude 1s the product of the current in m d the area of the loop and whose direction is the direction of the thumb as the fingars of the right hand follow the direction of the current. Comparison of E q (6-45) with the expression for the scalar electric potential of an electric dipole in Eq. (3-48),

reveals that, for the fwo cases, A is analogous to V. We call a small current-carrying , loop a magnetic dipole. The analogous quantities are as follows: .

- Electric Dipole Mngnet~c Dipole

, In a similar manner we can also rewrite Eq. (6-44) as


Except for the change of p to m, Eq. (6-48) has the same form as Eq. (3-49) does for the expression for E at a distant point of an electric dipole. Hence, the magnetic flux lines of a magnetic dipole lying in the xy-plane will have the same form as that of the electric field lines of an electric dipole positioned along the z-axis. These lines have been sketched as dashed lines in Fig. 3- 14. One essential difference is that the electric field lines of an electric dipole start from the positive charge + q and terminate on the negative charge -q, whereas the magnetic flux lines close upon themselves.+

6-5.1 Scalar Magnetic Potential

In a current-free region J = 0, Eq. (6-7) becomes

The magnetic flux density B is tfxm curl-free and can bc exprcssed as thc gradient of a scalar field. Let

B = -1'0 VT/;,,, (6-50)

where I/;, is called the scular. r~ltrgrretic polc~r~ti~il (esprcssed il~iqmperes). The negative sign in Eq. (6-50) is conventional (see the definition of the scalar electric potential V in Eq. 3-38), and the permeability of free space p, is simply a proportionality constant. Analogous to Eq. (3-40), we can write the scalar magnetic potential difference between two points. P2 and PI , in free space as

If there were magnetic charges with a volume density p, (A/m2) in a volume V', we would be able to find V, from

The magnetic flux density B could then be determined from Eq. (6-50). However, isolatcd magnetic charges have never been observed experimentally; they must be considered fictitious. Nevertheless, the consideration of fictitious magnetic charges in a mathematical (not physical) model is expedient both to the discussion of some magnetostatic relations in terms of our knowledge of electrostatics and to the estab- lishment of a bridge between the traditional magnetic-pole viewpoint of magnetism and the concept of microscopic circulating currents as sources of magnetism.

The magnetic field of a small bar magnet is the same as that of a magnetic dipole. This can bc vcrilied expcrimcntally by obscrving'thc contours of iron Jilings around a magnet. The traditional understanding is that the ends (the north and south poles)

QT

' Although the magnetic dipole in Example 6-7 was taken to be a circular loop, it can be shown (Problem P.6-13) that the same express~ons-Eqs. (6-45) and (6-48)-are obtalned when the loop has a rectangular shape, w~ th m = IS. as given in Eq. (6-46).

does for ~ a l c f ux .; that of :se lines that the :rminate ~selves.'

i o do)

dicnt of

(6- 5 0 )

; e g ; P : ; t e n d :on I ; (i~llcr-

(6- 5 1)

ume V',

(6-52) - owever, xust be xges in ,f some e estab- get ism

. n d i ~ , 4.

are\;- ' h POL-,

Problem &kngular

6-6 1 MAGNET' ITION AND EQUIVALENT CURRENT DENSITIES 213

. $ . a

of a magnet are the location of, respectively, positive and negative magnetic charges. For a bar mhgnet', the fictitious magnetic charges + q,, and - q,, are assumed to be separated by a pistance d and to form an equivalent magnetic dipole of moment

, ! m = q;d = a,lS. (6-53)

The scalar magnetic potential v,, caused by this magnetic dipole can then be found by following the procedure used in subsection 3-5.1 for findin, 0 the scalar electric potential that is caufed by an electric dipole. We obtain, as in Eq. (3-48),

II

Substitution of Eq. (6 -54) i n Eq. ( 6 5 0 ) yiclds thc same R as given in Eq. (6-48). We note that th,c ex~rcssions of the scalar magnetic potential V, , in Eq. ( 6 ~ 5 4 )

for a magnetic dipole are cxactly analogous to those for the scalar electric potential V in Eq. (6-47) for an electric dipole: the likeness between the vector magnetic potential A (in Eq. 6-35) and b' in Zq. (6-17) is nor as exact. Holvever, since magnetic ch:~rgcs do not exist in practical problems, I/;, rilust be determined from a givzn current distribution. This determination is usually not a simple process. Moreover. the curl- free nature of IZ indicatccl in Ey. (6--49), from which the scalar magnutic potential V,, is defined, holds only ~t points with 110 currcnts. In a region whcrc currents csis:. the magnetic field is rlot ~ c i l s ~ u ~ i i i c e , and,the scalar magnetic potential is not a single- valued function; hence tile magnetic potential difference evaluated by Eq. (6-51) depends on the path of il~tegration. For these reasons, we will use the circularing- current-and-vector-pbten;ial approach, instead of the fictitious magnetic-charge- and- scalar-potential approach. for the study of magnetic fields in magnetic materials. We ascribe the macroscopic properties of a bar magnet to circulating atomic currents (Ampkrian currents) caused by orbiting and spinning electrons.

6-6 MAGNETIZATION AN^ EQOIVALENT CURRENT DENSITIES

According to the elerhentary atomic mcdel of matter, all materials are composed of atoms, cach with a positively chargcd nuclcus and a numbcr of orbiting negatively charged electrons. The orbiting electrons cause circulating currents and form microscopic magnetic dipoles. In addition, both the electrons and the nucleus of an atom rotate (spin) on their own axes with certain magnetic dipole moments. The magnetic dipole moment. of a spinriing I I L I C ~ C ~ I S is u~u;tIly negligible compared to that of an orbiting or spinqing electron hecause of the much largcr nx~ss and lower anp1:ir velocity of the nucleds. A camplete understanditig of the magnetic effects of materials requires a knowledge of quantum mechanics. (We give a qualitative description of the behavior of different kinds of magnetic materials later in Section 6-9.)

In the absence of an external magnetic field, the magnetic dipoles of the atoms of most materials (except permanept magnets) have random orientations, resulting


in no net magnetic moment. The application of an external magnetic field causes both an alignment of the magnetic moments of the spinning electrons and an induced magnetic mom'ent due to a change in thc orbital motion of clcctrons. 111 ortlcr 10 obtain a formula for determining the quantitativc change in the magnetic llux dcnsity caused by the presence of a magnctic matcrisl. we kt in, hc thc ~nagaetic d ~ p u l s moment of an atom. If there are H atoms per unit volun~e, we define a inugi~etizution vector, M, as

k = l M = lim - A W O AV (A/m) 9

which is the volume density of magn'etic dipole moment. The magnetic dipole moment dm of an elemental volume dti' is'dm = XI da' that. according to Eq. (6-451, will produce a vector magnetic potential

Using Eq. (3-75). we can write Eq. ( 6 - 5 6 ) a s

Thus,

where V' is the volume of the magnetized material. We now use the vector identity in Eq. (6-28) to write .

and expand the right side of Eg. (6-57)hto two terms:

The following vector identity (see Problem P. 6- 14) enables us to change ;he volume integral of the curl of a vector into a surfacc integral.

jv, P x F d d = -$s, F x ds', (6- 60)

where F is any vector with continuous first derivatives. We have, from Eq. (6-59)

\es both induced srdcr to . density ;. dipole 'ti:ation

d

(6-55)

noment Is), will

16 56)

f-

volume

P (6--60)

6-6 1 MAGNETIZATION AND E ~ I V A L ~ N T CURRENT QENSITIES 21 5

I

. where a: is the unit outward normal vector from ds' and St is thc surface bounding the volume V'. I, ,. ,' I

A comparison ofthe expressions on the right side of Eq. (6-61) with the form of A in Eq. (6-22), expdssed in terms of volume current density J suggests that the effect of the magnetizatio$vectqr is equivalent to b&th a volume current density

I

and a surface current density

J,, = M x a,, (Ajm). (6 -63) ,

In Eqs. (6-62) and (6-6.i) we hare omimd the primes on V and a,, for simplicity, since i t is clear that both refer to the coordinates of the source point where the map- nctization vector M exisis However. the primes should be retained when there is n possibility of confusing tile coordinates cf the source and field points.

The problem of finding the magnetic flux density E caused by a given iaiume . density of magnetic dipvls moment i\.i is then reduced to finding the equiiaicnt

tnaqnerizoiion current iie.&rirs J, and J,,,, by using Eqs (6-61) m d (6-631, determining A from Eq. (6-611, and then obtaining B from the curl of A. The externally applied magnetic field. if i t also exists, must be accounted for separately.

The mathematical derivation of Eqs (6-62) and (6-63) is straightforivord. Tbs equivalence of a volume density of magnetic dipole moment to a volume current density and a surface cuxent density can be appreciated qualitatively by rekrring to Fig: 6-9 where a cross scction of a magnetized material is shown. It is assumed that an externally applied magnetic field has caused the atomic circulating currents to align with it, thereby magnetizing the material. The strength of this magnetizing

Fig. 6-9 A cross section of a magnetized material.


effect is measured by the magnetization vector M. On the surface of the material, there will be a surface current density J,,, whose direction is correctly given by that of the cross product M x a,. If M is uniform inside the material, the currents of the neighboring atomic dipoles that flow in opposite directions will cancel everywhere, leaving no net currents in the interior. This is predicted by Eq. (6-62), since the space derivatives (and therefore the curl) of a constant M vanish. However, if M has space variations, the internal atomic currents do not completely cancel, .esulting in a net volume current density J,. It is possible to justify the quantitative relationships between M and the current densities by deriving the atomic currents on the surface a i d in the interior. But as this additional derivation is really not necessary and tends to be tedious, we will not attempt it here.

Example 6-8 Determine the magnetic flux density on the axis of a uniformly magnetized circular cylinder of a magnetic material. The cylinder h is a radius h, length L, and axial magnetization M.

Solution: In this problem concerning a cylindrical bar rnagqt. let the axis of the magnetized cylinder coincide with the z-axis of a cylindrical~oordinatc system, as shown in Fig. 6-10. Since the magnetization 1CI is a constant within the magnet, J, = V' x M = 0, and there is no equivalcl;t volume current density. The cquivulent magnetization surface current density on the side wall is

Jms = M x a; = (aJ4) x a,

= a,M. (6-64)

The lnagrlet is then like a cylindrical sheet with a lined clrrrerlt density of M (A/m). There is no surface current on the top and bottom faces. In order to find B at P(0, 0, s), we consider a differential length dz' with a current a,M dz' and use Eq. (6-38) to

Fig. 6-10 . A uniformly magnetized circular cylinder (Example 6-8).

atcrial, by that

of the where, e space s space ~g In a mhips surface 3 tends

formly dius b,

, o i the 3 . ~ 5 t i r

.?asnet, ! I \ d '

i h 64)

( A m). (0, 0, I), -38) to

n,

obtain , j

, i l o ~ b 2 dzl dB = a, -- 2[(z - z'JZ + b 2 F

and , , +,

' B = - J ~ B ' = a, J L 0 p o ~ b 2 dz'

2 [ ( z - z')' + b2] 3 / 2

6-7 MAGNETIC FIELD INTENSITY AND RELATIVE PERMEABILITY

Because the application af an txternal magnetic fieid ciloses both :In :~ligilmeor ai the internal dipole moments and an induced magnetic moment in 3 magnetic material. we expcct that the resultant magnetic flux density in the presence of Y magnetic material ~ i i l be different from its value in free space. The n~acroscopic etTect a i magnetization can be studied by incorporating the equivalent volume current densit!. J, in Eq. (6-62). into the basic curl equation, Eq. (6-7). We have

We now define a new f~mdamental field quantity, the inagnetic field intensity, H , such that

The use of the vector H enables us to write a curl equation relatlng the magnetic field and the distribthon of free'currents in any medium. There is no need to deal explicitly with the ninpnetization vector M or the equivalent volume current density .I,,,. ~ o m h i i i T i ~ ~ Eqs. ((660) and 36-67). we ohlain the ncw cqu:ition

(6-68)

where J (A/m2) is the volume density olfiee ci~rlrnt. Equations (6-6) and 16-68) are the two fundamental governing differential equations for magnetostatics in any


medium. The permeability of free space, pO, does not appear explicitly in these two equations.

The corresponding integral form of Eq. (6-68) is obtained by taking the scalar surface integral of both sides.

or, according to Stokes's theorem,

where C is the contour (closed path) bounding the surface S, and 1 is the total current passing through S. The relative diiections of C and current flow I follow the right- hand rule. Equation (6-70) is another form of Ampc'.re's circuital law: It states that the circulation of the ntugneticjeld intensity urounil any closed path is equal to the J rcc c l i l - r e n r jlor\.inc/ T / I I . O L I ~ / ~ ~ / I C S I I I . /~KC' h l l l 7 d d I.!. p ~ l t . / ! ~ ~ - \ s \ve iildic,ltcd in Section 6-3. Amptre's circuital law is most useful in determi'ning the magnetlc field caused by a current when cylindrical symmetry exists-that is, when there is a closed path around the current over whichthe magnetic field is constant.

When the magnetic properties of the medium are linear and isotropic, the magnetization is directly proportional to the magnetic field intensity:

where zm is a dimensionless quantity called magnetic susceptibility. Substitution of Eq. (6-71) in Eq. (6-67) yields

(6-72b)

where

is another dimensionless quantity known as the relative permeability of the medium.

b <

. ) L

I

6-7 / M A ~ N E T ~ C FIELD INTENSITY AND RELATIVE PERMEABILITY 219

Fig. 6-11 Coil on air gap (Example 6-

ferromagnetic toroid w ~ t h -9).

The parameter p = u,p. is the absoliite perazeu8ilit)~ (or, sometimes. just pemiubil i t j 1 of the medium and is measured in H/m; z,,,, and therefore p,, can be a function of space coordinarcs. For a simple mcdium - linear, isotropic, anct homogeneous --

z,,, m d p, are consttlnts. The permeability of nost materials is very close to that of free space (p , , ) . Fdr

ferromagnetic materials such as iron. nickel, and cobalt. p, could be very large (50-5000, and up to 106 or more for special alloys): the permeability depends no[ on ly on lhc magnitude ol 11 but also on liic prcvious hibtory of the material. Section 6-8 contains some quali~ative discussions of the macroscopic behavior of magnetic materials.

Examplc 6-9 Assunw illat iZl turns ol'wire arc wo~inci around a roroid~ll care ol'.i ferromagnetic material with permeabiilty p. The core has a mean radius r,], 3 circular cross section of radiUs u iu I< r,,), and a narrow air gap of length I,, as shown in Fig. 6-11. A steady currdnt I,, Rows in the wire. Determine (a) the magnetic flux denhity.

B f , in the ferromagnetic core; (b) the magnetic field intensity, H,., in the core; and (c) the magnetic field intensity, H,, in the air gap.

Solution

a) Applying Ampkke's circuital law, Eq. (6-70), around the circular contour C, which has a mean radius r,, we have

lilliir lr:~k:~ga is ncg l c~k~ i . i l l r S : I ~ C w h l 11~1s will lloa iii 170111 ilia Scrron~ayctic corf ;md in ihc air g:lp: I f lhc fringing clkul ol. lllc llur i n the air gap is also ncg- lected, the magnetic Hux density B in both the core and the air gdp will also be

. the same. However, Zccause of the different permeabilities, the magnetic field intensities in both parts will be different. We have

* 220 STATIC MAGNETIC FIELDS / 6

In the ferromagnetic core,

and, in the,air gap,

Po

Substituting Eqs. (6-75), (6-76), and (6-77) in Eq. (6-74), we obtain

and

B I U O P N I ~ ; ' po(2ni-,, - + pi,

'

b) From Eqs. (6-76) and (6-78) we get

c) Similarly, from Eqs. (6-77) and (6-78). wc havc

Since H,/H, = p/p0, the magnetic field intensity in the air gap is much stronger than that in the ferromagnetic core.

Why do you think the condition a cc r, is stipulated in this problem?

6-8 MAGNETIC CIRCUITS

The problem in Example 6-9 is, essentially, one of a magnetic circuit in which the current applied to the winding causes a magnetic flux to flow.in the ferromagnetic core and thc air gap in series. We define the line integral of magnetic field intensity around a closed path,

as magnetomotiw force,* mmf. Its SI unit is ampere (A): but, because of Eq. (6-74), mmf is frequently measured in ampere-turns (A t). An mmf is not a force measured in newtons.

Assume 1'; = N1,denotes a magnetomotive force that causes a magnetic flux, 0, to flow in a magnetic circuit. If the radius of the cross section of the core is much

' Also called magnetomotance.

r 6 -76)

(6-77)

(6-78)

i 0 -,Pl

( 6 SO)

.rongcr

ich the lgnetic .tensity

( 6 p, ' J X I I , , J

x Hu. s much

I L

: 6-8 1 MAGNETIC CIRCUITS 221

. : smaller than the mean radius of the toroid, the magnetic flux density B in the core is approximately co&ant,'and

cf, E B S , (6-81)

wlicrc S is thc cross-scclionnl area of the cute. Combination of Eqs. (6-81) and (6-78) yields 4

Equation (6-82) carfbe rewritten

with

where tl = 2nr, - fg is the length of the ferrorhagnetic core, and

Both 3, and 2, have the rime form :is the formula. Lq. (5-1 3) . for ihc DC res~st:incc - .

o 1 s : p i c f 1 o 1 1 c 1 o s i i I a i cross s i S. U l l h arc called rcl i iaunr~: 2 , of the fcrromiignctic corc; and r?,,, of the ;iir g ip . Tile SI unit for reluctance is reciprocal henry (H- I ) . The fact that Eqs (6-84) and 16-85] are as they are, even though the core is not straight, is a consequence of assuming that B is approximately constant over the core cross section.

Equation (6-83) is analogous to the expression for the current I in an electric circuit, in which an idealvoltage source of emf Y is connected in series with two resistances Rf and R,:

(a) Magnctic circuit, (b) Elcctric circuit.

Fig. 6-12 Equivalent magnetic circuit and analogous electric circuit for toroidal coil with air gap in Fig. 6- 1 1.

222 STATIC MAGNETIC FIELDS / 6

, ,

respectively. Magnetic circuits can, by analogy, be analyzed by the same techniques . we have used in analyzing electric circuits. The analogous quantities are

Magnetic Circuits Electric Circuits

mmf, YC,, (= N I ) emf, -/'

magnetic flux, 0 electric current, I reluctance, 94 resistance, R

permeability, p conductivity, a

In spite of this convenient likeness, an exact analysis of magnetic circuits is inherently very difficult to achieve.

First, it is very difficult to account for leakage fluxes, fluxes that stray or leak from the main flux paths of a magnetic circuit. For the toroidal coil in Fig. 6-11, leakage flux paths encircle every turn of the winding; they .partially transverse the space around the core, as illustrated, because the perrneabilit<of air is not zero. (There is little need for considering leakage currents outside the conducting paths of electric circuits that carry direct currents. The reason is that the conductivity of air is practically zero compared to that of a good conductor.)

A second difficulty is the fringing effect that causes the magnetic flux lines at the air gap to spread and bulge.+ (The purpose of specifying the "narrow air gap" in Example 6-9 was to minimize this fringing effect.)

' A third difficulty is that the permeability of ferromagnetic materials is dependent on the magnetic field intensity; that is, B and H have a nonlinear relationship. (They may not even be in the same direction.) The problem of Example 6-9, which assumes a given p before either B, or H, is known, is therefore not a realistic one.

In a practical problem, the B-H curve of the ferromagnetic material, such as that shown later in Fig. 6-15, should be given. The ratio of B to H is obviously not a constant, and Bf can be known only when H f is known. So how does one solve the problem? Two conditions must be satisfied. First: the sum of H,L, and H f L f must equal the total mmf N I , :

H J g + H,.tJ = NNI,. (6-87a)

' In order to ohtain a more accurate numerical result, it is customary to consider the elfcctive area of the air gap ;IS \Iighlly Ixrgcr th:111 the c ross -scc t io~~~~i : t r u o r tllc l ' c r r i ~ ~ ~ l ~ ~ g r ~ c ~ i c CON, wit11 c;~ch of [ I I C 1inc;d dimensions oC the core cross section increased by the length of 111e air gap. if we wcre to make n correction

. .. like this in Eq. (6-75). 5, would become

~niques

x1ts 1s

)r leak 0 1 1 , \c I*

: / C L p'lt'

v r t y ...

ar the ,~p" In

rridcnt (They >u1ncs

ich as J y not : sohe

H, 1~

4 7 a )

/1 a of the r 11nc. '

rrecqo. _

6. h . 1 MAGNETIC ClRCUlTS 223

. 4

Second, if we asstime no leakage flux, the total flux CD in the ferromagnetic core and in the air gap must be the same, or B, = B,,:+

\

Substitution of Eq. (6-87b) in Eq. (6-87a) yields an equation relating BJ and H, in the core:

This is an equatio$fbr'k straight line in the B-H plane with a negative slope- ,u,C ,/&,. The intersection of this line and the given B-H curve determines the operating point. Once the operating has been found, p and H , and all other quantities can be obtained.

The similarity between Eqs. (6-83) and (6-86) can be extended to the writing of two basic equations for magnetic circuits that correspond to Kirchhoff's voltage and current laws for electr~c circuits. Simi;nr to Kirchhofl's voltage law in Eq. (5-37), we may write, for any closed path in a magnetic circuit,

Equation (6-89) states that around a closed p u ~ h in a magnetic circuir rhe algebraic s u m of umpere-turns is e q t d to the algebraic sum of the products of the re2ucti;nce.s und fluxes.

Kirchhoff's current lsw for a junction in an electric circuit, Eq. (5-33), is a consequence of V . J = 0. Similarly, the fundamental postulate V . B = 0 in Eq. ( 6 4 ) leads to Eq. (6-8). Thus we have

which states that the dgebraic sum of all the magneticJuxes jlowing out of a juncrion in a magnetic circuit is zero. Equations (6-89) and (6-90) form the bases for. respectively, the loop and ndde, analysis, of magnetic circuits.

Example 6-10 Consider the magnetic circuit in Fig. 6-13(a). Steady currents I, and I, flow%Twindings of, respectively; N , and N 2 turns on the outside legs of the

' T h ~ s assumes an equal crobs-sectional area for the core and the gap. If the care were to be constructed of ihsulated laminations of ferromagnetic material, the effective area for flux passage in the core would be smaller than the geometrical cross-sectional area, and B, would be larger than B, by a factor. This factor can be determined from tho datq on the insulated laminations.

1 224 STATIC MAGNETIC FIELDS / 6

(a) Magnetic core with current-carrying windings. (b) Magnetic circuit for loop analysis.

Fig. 6-13 A magnetic circuit (Example 6-10).

ferromagnetic core. The core has a cross-sectional area Sc and a permeability p. Determine the magnetic flux in the center leg.

Solurioiz: The equivalent magnetic circoit Sor loop ;mllyis is.sIlown in Fig. b- l3lb). Two sources of mmf's. N , I l and N 2 1 2 . arc shown with p r o p ~ ~ o l : l r i t i c s is scrics with reluctances A, and .f12 r ~ s p ~ ~ I i v c I y . This is t i l~vioilsl~ ii two-iaop I I C ~ \ V ( > I . ~ . Since wc arc determining mapoc~ia llua i ~ i the center leg I',P,, i t is crpediellt to choose the two loops in such a way that only one loop flux (a),) flows through the center leg. The reluctances are computed on the basis of average path lengths. These are, of course, approximations. We have

The two loop equations are, from Eq. (6-89).

Loop 1 : N I I l = (!MI + g 3 ) ( b 1 + LVlO2, (6-32)

Loop 2: N I I l - N212 = g l Q l + ( 9 1 + g2)02. (6-93)

Solving these simultaneous equations, we obtain

which is the desired answer.

6-9 1 B E H A V ~ R OF MAGNETIC MATERIALS 225 4

. Actually sincd t ,e magnetic fluxes and thhrefore the magnetic flux densities in the three legstare ! ifferent, different permeabilities should be used in computing the reluctances in E$s. (6--!,la), (6-91b) and (8-91c). But thc value of permeab~llty, in turn, depends dn the magnetlc flux density. The only way to improve the accuracy of the solution, prokided thC B-H curve of the core material IS given, is to use procedure of ducces8ive approximation. For instance, Q,, Q2, and cD, (and therefore B,, B2 , and B3) are first solved with an assumed p and reluctances comp~~ted from the three parts of Eq! (6-91). From B, . B,, anQ B, the corresponding ,ul, p 2 , and p , can bc found from'lthe B-H curve. These will modify the reluctances. A second approximation for dl, B2; and B , is then obtained with the modified reluctances. From the new flux dgnsities, new permeabilities and new reluctances are determined This procedure is repeated until further iterations brmg little changes in the computed values.

6-9 BEHAVIOR OF MAGNETiC MATERIALS

In Eq. (6-71), Section 6 l 7 , we dsscribed the macroscopic magnetic property of i linear, isotropic med;lum by defining the magnetic susceptibility x,,, a dimensionless coefficient of proportionality between magnetization M and magnetic field intensity H. The relative pernleabiljty p, is simply 1 + x,. Magnetic materials can be roughly classified into three main.groups in accordance with their ,u, values. A materiai is said to be

Diamagnetic, if p, 5 I :(.y, is a very small negative number).

Purarnuyneric, if ,u, 2 L,(x, is a very small positive number) Ferromugnetic, if p, >,:I (2, is a large positive number).

. . As mentioned before, a thorough understanding of microscopic magnetic phenomena requires a knowledge of quantum mechanics. In the following we give a qualitative description of the behavior of the various types of magnetic materials based on the classical atomic modkl.

In 3 cli(i~ti(l{]l!~ficmatcri:ll thc nct magnclic ~~iomcnt due to thc orbital and spinning motions of the electrons in any particular atom is zero in the absence of an externally applied magnetic field. As predicted by Eq. (6-4), the application of an external magnetic field to this material produces a force on the orbiting electrons, causing a perturbation in the angular velckities. As a consequence, a net magnetic moment - ,

is created. This is a hrocess of induced magnetization. According to Lenz's law of electromagfie'etic induction (Section 7.-2), the induced magnetic moment always opposes the applied field, thus reducing the magnetic flux density. The macroscopic effect of this process-is equivalent to that of a negative magnetization that can be described by a negative rnagnetic susceptibility. This effect is usually very small, and x,,, for most known diamagnctic materials (bismuth, copper, lead, mercury, germanium, silver, gold, diam.ond) is in the order of -


Diamagnetism arises mainly from the orbital motion of the electrons within an atom and is present in all materials. In most materials it is too weak to be of any practical importance. The diamagnetic effect is masked in paramagnetic and ferromagnetic materials. Diamagnetic materials exhibit no permanent magnetism, and the induced magnetic moment disappears when the applied field is withdrawn.

In some materials the magnetic moments due to the orbiting and spinning electrons do not cancel completely, and the atoms and molecules have a net average magnetic moment. An externally applied magnetic field, in addition to causing a very weak diamagnetic effect, tends to align thc molecular magnctic moments in tlzc direction of the applied field, thus increasing the magnetic flux density. The macroscopic effect is, then, equivalent to that of a positive magnetization that is described by a positive magnetic susceptibility. The alignment process is, however, impeded by the forces of random thermal vibrations. There is little coherent interaction and the increase in magnetic flux density IS quite small. Materials with this behavior are said lo bc />tn.trrlrtr!/r~c,/ic.. P ;~ r ;~~n ;~p ic l i c m;ltcri;~ls p c r a l l y haw vcry slnall positive v;~Iucs 01' 111;1g11clic s~~sccplilGlily. i l l lhc 01dcr ot I0 ' ; ~ I u t l l i ~ l u ~ l l , 111;1$11csiutll. titanium, and tungsten. - --.

Paramagnetism arises mainly from the magnetic dipole moments of the spinning electrons. The alignment forces, acting upon molecular dipoles by the applied field, are counteracted by the deranging effects of thermal agitation. Unlike diamagnetism. which is essentially independent of temperature, the paramagnetic elrect is temperature dependent, being stronger at lower temperatures where there is less thermal collision. . The magnetization of ferromagnetic materials can be many orders of magnitude larger than that of paramagnetic substances. (See Appendix B-5 for typical values of relative permittivity.) Ferromcrgnctism can be explained in terms of magnetized tlort~rrirr.~. According l o this ~l~otlcl, w i ~ i c l ~ 11:~s I x x r i cxpcrit~\cnt:~lly co~l(i~.ll~ctl, ;I

ferromagnetic mutcrial (such as cobalt, nickel, and iron) is composcd of many small domains, their linear dimensions ranging Srbm a few microns to about 1 mm. 'These domains, each containing about 10'"r 1016 atoms, are fully magnetized in the sense that they contain aligned magnetic dipoles resulting from spinning electrons even in the absence of an applied magnetic field. Quantum theory asserts that strong coupling forces exist between the magnetic dipole moments of the atoms in a domain, holding the dipole moments in parallel. Between adjacent domains there is a transition region about 100 atoms thick called a clonzailz wall. In an unmagnetized state, the magnetic moments of the adjacent domains in a ferromagnetic material have different directions, as exemplified in Fig. 6-14 by t h e polycrystalline specimen shown. Viewed as a-whole, the random nature of the o;ientations in the various domains results in no net magnetization.

, . ,. When an external magnetic field is applied to a ferromagnetic material, the walls

. . of those domains having magnetic moments aligned with the applied field move in such a way as to make the volumes of those domains grow at the expense of other

thin an of any

3 ferro- :m, and sn. pinning :vcrage lislng a :s in the macro: wrihed xpeded oil and /lor ;ire

,mji\ive ;:csium,

.\!nlF'- si f i c .1~ .

net! 1s tcm- i ! l c ~ r ~ > ~ l

gnitude ! values metized med, a iy small I. These .le sense ns everi

strong hmain, ansition .ate, the liirerent s h q

:om,,,ls

hc \va. . :nave in of other

6-9 1 BEHAVIOR OF MAGNETIC MATERIALS 2?? ,

Magnetized -domain Domain

-wall

1:iy. 0-1 J Do~nain S L S L I C ~ L I I - C 01' a polycrystalline ferromagnetic specimen

domains. As a result magnetic flux density is increased. For weak applied fields, say up to point P , in t ig . 5-15, domain-wall movements are reversible. But when an applied field becorned stronger (past P,). domain-wall movements are no !onger reversible. and domah n)tation towitrd the direction of thc irpplicd ficid will rdso O ~ L I ~ . For cr;wpic, if all ipplicil licld is rcduced lo zero at point I ) , , the B-H rcla- tionship will not foliop th: solid curve P,P,O, hut will go down from P , to P;. ;iionp the iines of the broken curve in the ligure. This phenomenon of magnetization lagzing behind the field producing it is called A?;stcizsis, which is derived from a Greek word meaning "to lag." 4 s ihe applied field becomes even much stronqer (past P I to P,). domain-wall motion and domain rotation will cause essentially a total ah, nnment of the microscopic magnaic moments with the applied field. at which point the magnetic material is said to have reached rorilrcsioii. The curve O P , P,P, on the R I 1 plunc is callcd thc irc~i.t,rtrl mr~picliztilioti clrl-vc.

'

If +he applied magnetic field is reduced to zero from the viilue at P,. the magnetic flux density does not go to zero but assumes the value at B,. This value is called the residaal or reinanent flux &i?sir~~ (in W b p ' ) and is dependent on the maximum applied field intensity. The existence of a remanent flux density in a ferromagnetic material makes permahent magnets possible.

Fig.6-15 Hysteresis loops in B-H plane for ferromagrletic material. '


In order to make the magnetic flux density of a specimen zero, it"is necessary to apply a magnetic field intensity H , in the opposite direction. This required H , is called coersive force, but a more appropriate name is coersivefield intensity (in Aim). Like B,, H , also depends on the maximum value of the applied magnetic field intensity.

It is evident from Fig. 6-15 that the B-H relationship for a ferromagnetic material is nonlinear. Hence, if we write B = p H as in Eq. (6-72a), the permeability p itself is a function of the magnitude of H. Permeability ,LL also depends on the history of the material's magnetization, since-even for the same H- we must know the location of the operating point on a particular branch of a particular hysteresis loop in order to determine the value of hi exactly. In some applications a small alternating current may be superimposed on a large steady magnetizing current. The steady magnetizing ficld intensity locates the operating point. and the local slope of the hysteresis curve at the operatins point determines the incre~ilentcll prrnwahilit!..

Ferromagnetic materials for use in electric generators, motors, and transformers should have a large magnetization for a very small applied field: they should have tall :~nd narrow Iiystercsis loops. As the applied magnctic ficld intensity varics pcrio~lically hciwcc~l 5 11,,,: , , . the I~yslcluis loop is \raced w_cc per cycle. 'l'llc ;Ire:\ of the hysteresis loop corrtxponds to energy loss (h!~sr~w.si.s los.;jper unit volume pcr cycle (Problem P. 6-21). Hysteresis loss i? the energy lost in the form of heat in over- coming the friction encountered during domain-wall motion and domain rotation. Ferromagnetic materials, which have tall, narrow hysteresis loops with small loop areas, are referred to as "soft" materials; they are usually well-annealed materials with very few dislocations and impurities so that the domain walls can move easily.

Good permanent magnets, on the other hand, should show a high resistance to demagnetization. This requires that they be made with materials that have large coercive field intensities H , and, hence, fat hysteresis loops. These materials are referred to as "hard" ferromagnetic materials. The coercive field intensity of hard ferromagnetic materials (such as Alnico alloys) can be lo5 (A/m) or more, whereas that for soft materials is usually 50 (A/m) or less.

As indicated before, ferromagnetism is the result ofstrong coupling effects between the magnetic dipole moments of the atoms in a domain. Figure 6-16(a) depicts the atomic spin structure of a ferromagnetic material..When the temperature of a ferromagnetic material'is raised to such an extent that the thermal energy exceeds the couplicg energy, the magnetized domains become disorganized. Above this critical temperature, known as the curie temperature, a ferromagnetic material behaves like a paramagnetic substance. Hence, when a permanent magnet is heated above its curie temperature it loses its magnetization. The, curie temperature of most ferromagnetic materials lies between a few hundred to a thousand degrees Celsius, that of iron being 770°C.

2 : . .. .. *:. < a , , .. Some elements, such as chromium and manganese, which i r e close to ferro-

magnetic elements in atomic number and arc neighbors of iron in the pcriodic table, also have strong coupling forces between the atomic magnetic dipole moments;

6-9 / BEHAWOR OF MAGNETIC MATERIALS . '9

1:

I y i ~ . 6-16 Schcm:llic atomic spin struclurcs I'or la) ferromagnetic, (b) mtikrromagnetic, and ic) fcrrimagnctic materials.

but their coupling forces produce antiparailel alignmen~s of electron spins. 2s illustrated in Fig. 6--16(b). The s$ns alternaic in direction from atom to atom and rcsult in no net magnetic moment. A material possessing this Eroperty is said to be m r i - ferrontagnetic. ~ntiferronla~netism is also temperature dependent. When an anti- ferromagnetic material is h a t e d above its curie temperature, the spin directions suddenly become random and the material becomes paramagnetic.

There is another class of inagnetic materials that exhibit a behavior between ferromagnetism and antiferromagnetism. Here quantum mechanical effects make the directions of the magnetic moments in the ordered spin structure ‘1 , 1 ternate and the magnitudes unequal, resulting in a net nonzero magnetic moment, as depicted in Fig. 6-16(c). These materials are said to be ferrimagnetic. Because of the partial cancellation, the maximum magnetic flux density attained in a ferrimagnetic substance is substantially lower than that in a ferromagnetic specimen. Typically, it is about 0.3 Wb/m< approximately one-tenth that for ferromagnetic substances.

Fcrrites are a subgroup of fcrrimagnetic material. One type of ferrites, called ~ i ~ u y ~ l e r i c spinels, crystaliize in a complicated spinel strucure and I1;~ve the formula XO . Fe203, where denotes a divalent metallic ion such as Fe, Co, Ni, Mn, Mg, Zn, Cd, etc. These are ceramic-like compounds with very low conductivities, (for instance, lO-%d_(S/m) compared with lo7 (S/m) for iron). Low conductivity limits eddy-current losses at High frequencies. Hence ferrites find extensive uses in such high-frequency and microwave :ipplicntions as corcs for FM antennas, high-frequency trnnsformurs, and phase shii'lcrs. Other fcrritcs include magnetic-oxide garnets, of which Yttrium-Iron-Garnet ("YIG," Y,Fe,D,,) is typical. Garnets are used in microwave multiport jut~ctions. . '

4 230 STATIC MAGNETIC FIELDS I 6 I

I I

I / 1 .

I 6-10 BOUNDARY CONDITIONS FOR I MAGNETOSTATIC FIELDS

In order to solve problems concerning magnetic fields in regions having media with different physical properties, it is necessary to study the condition: (boundary conditions) that B and H vectors must satisfy at the interfaces of different media. Using techniques similar to those employed in Section 3-9 to obtain the boundary conditions for electrostatic fields, we derive magnetostatic boundary conditions by applying the two fundamental governing equations, Eqs. (6-6) and (6-68), respectively, to a small pillbox and a small closed path which include the interface. From the divergenceless nature of the B field in Eq. (6-6), V B = 0, we may conclude directly, in light of past experience, that the normal component of B is continuous across an interface; that is,

For linear media, B, = p1Hl and B2 = j1,H2, Eq. (6-95) becomes 1-

(6-96)

The boundary condition for the tangential components of magnetostatic field is obtained from the integral form of the curl equation for H, Eq. (6-70), which is repeated here for convenience:

We now choose the closed path rbcda in Fig. 6-17 as the contour C. Applying Eq. (6-97) and letting bc = da = Ah approach zero, we have

where J,, is the surface current density on the interface normal to the contour C. The direction of J , is that of the thumb when the fingers of the right hand follow

Fig. 6-17 Closed path about the interface "

of two media for determining the boundary condition of H,.

dia with lry con- . 3. Using ,

iry con- ~ o n s by rc\r,cc-

c. From onclude nrinuous

(6-98)

xbur C. d follow

P

rface ldary

6 -10 ~ O U N D A R Y CONDITIONS COR MAGNETOSTATIC FIELDS 231 : I G

? the direction of the &h. In Fig. 6-17, the positive direction of J," for the chosen path is out of the paper. ~ i l e following is a more concise expression of the boundary

+

, condition for the tangential components of H, which includes both magnitude and direction relations (Problem P. 6-22).

a,l2 x ( H I - H,) = J, (A/m), (6-39)

whom a,,, is the olrlwut+ uililnortt~sl j,l)rtt t ~ ~ a d i u n ~ 2 :it thc in~crf:lcc. Thus, the tongcntiiri ... , component oj the HJi~ ld is discontis~u,u.s across un interjoce where o surjbce cilrrerlt exists, the amount of.disccntinuity being determined by Eq. (6-99).

When the condu~kvities of both media i r e finite, currents are defined by voiume current densities a n d free surface currents do not exist on the interface. Hence. J, equals zero, and the tut.qenrio1 conlponcnt of H is continuous ucioss the boiiridury h . . ' . . o 1 h y i c I ~ I ~ I : 11 1s discontinuoos only whcn an intcrfacc with an ideal pcrfcct conductor or a ~up~xconductor is assumed.

i

Example 6-11 Two magnetic media with permeabilities ji, and p , have a common boundary, as shown in FIG. 6-18, The magnetic field intensity in medium 1 at ;hi. point l', has a magnihde 11, and m:~kcs an angle a , with the normal. Determine he magnitude and the directic.n of the magnetic field intensity at point P , in medium 1.

Solution: The desired un1,ooyn qu:intities 'lrc Hz and a,. Continuity of the normal component of B field requires, from Eq. (6-96),

p2H2 cos z2 = p l H 1 cos a, . (6-1OC)

Since neither of the medic is a perfect conductor, the tangential component of H field is continuous. We have

H , sin a , = H 1 sin a l .

Division of Eq. (6-101) by Eq. (6-100) gives

+ Fig. 6-18 Boundary conditions for magnerostatic field at an interface (Example 6-1 1).

. .- 232 STATIC MAGNETIC FIELDS 1 6

which describes the refraction property of the magnetic field. The magnitude of H z is

H 2 = ,/- = , /(H~ sin,^,)^ + ( H z cos a?)'

We make three remarks here. First, Eqs. (6-102) and (6-104) are entirely similar to. respectively. Eqs. (3-119) and (3-120) for the electric fields in dielectric media- except for the use of permeabilit'ies (instead of permittivities) in the case of magnetic fields. Second, if medium 1 is nonmagnetic (like air) and medium 2 is feriomagnetic (like iron), then ~ 1 , >> p, and. from Eq. (6-102), a, will be nearly ninety degrees. Thw means that for any arbitrary angle a, that is not close to zero. the magnetic field in a ferromagnetic n~ed iun~ runs almost parallel to thc intc;fa~c: Third, if medium I is ferromagnetic and medium 2 is air (p, >> p2), then a, will be nearly zero; that is. if a magnetic field originates in a ferromaghetic medium, the flux lines will emerge into air in a direction almost normal to the interface.

In current-free regions the magnetic flux density B is irrotational and can be expressed as the gradient of a scalar magnetic potential Vm, as indicated in Sec- tion 6-5.1.

B = -$i'V,. (6-105)

Assuming a constant p, substitution of Eq. (6-105) in V B = 0 (Eq. 6-6) yields a Laplace's equation in Vm:

VZVm = 0. (6-106)

Equation (6-106) is entirely similar to the Laplace's equation, Eq. (4-lo), for the scalar electric potential V in a charge-free region. That the solution for Eq. (6-106) satisfying given boundary conditions is unique can be proved in the same way as for Eq. (4-10)-see Section 4-3. Thus the tcchniqucs (method of images and method of separation of variables) discussed in Chapter 4 for solving electrostatic boundary- value problems can be adapted to solving analogous magnetostatic boundary-value problems. However, although electrostatic problems with conducting boundaries maintained at fixed potentials occur quite often in practice, analogous magnetostatic problems with constant magnetic-potential boundaries are of little practical importance. (We recall that isolated magnetic charges do not exist and that magnetic flux lines always form closed paths.) The nonlinearity in the relationship between B and H in ferromagnetic materials also complicates the analytical solution of boundary- value problems in magnetostatics.

. , . " I I 6-11 1 INO~CTANCES AND INDUCTORS 233

i ?

i I . t : $ \

6-11 INDUCTANCES AND I ~ D U C T ~ R S f !

(6-103) Consider two neighboring closed loops, C, add C, bounding surfaces S l and S2 i respectively, as shown in Fig. 6-19. If a current I, flows in C,, a magnetic field B,

: ofH, is will be created. Some of the magnetic flux due to B, will link with C , - that is, wlll pass through the surf$ce S2 bounded by C,. Let us designate this mutual flux a,,. We have

ecs. This i

? can be i in Sec-

1, for the . (6-106) :ty as for elhod of ~undary-

:ry-'PY unci ,s

.etostafic 11 imf xtic flux :n B and mndary-

From physics we knok thiit a time-varying I , (and therefore a time-varying 81,) wll produce an induced e~ectromotive force or voltage in C2 as a result of Foraday's low of electromagnetic i$uction. (We defer the discussion of Faraday's law untll the ,

next chapter.) I Iowcdor, ( I , , , cxists cvcn ~f I , is a steady DC current. From ~ i o t - ~ a v a r ! law, Eq. (6-31), we see that B , is directly proportional to I, ;

hence ( D l , is also proportional to I,. We write

where the proportionslity constant L,, is called the mutual inductance between loops C , and C,, with SI unit henry (H). In case C2 has N , turns, thejux linkage A,, due to Q 1 2 is

A12=N2@12 (Wb), and Eq. (6-108) generalizes to

Fig. 6-19 Two magnetically coupled loops.


j ! '

The mutual inductance between two circuits is then the magneticjux linkage with one circuit per unit current in the other. In Eq. (6-108), it is implied that the permeability of the medium does not change with I , . In other words, Eq. (6-108) and hence Eq. (6-11 1) apply only to linear media. A more general definition for L, , is f t v

Some of the magnetic flux produced by I , links only with C, itself, and not with C,. The total flux linkage with C, caused by I , is

!

1

The self-inductance of loop C, is defined as the magneticflux linkage per: unit current in the loop itself; that is,

! L

for a linear medium. In general,

,

The self-inductance of a loop or circuit depends on the geometrical shape and the 1 f physical arrangement of the conductor constituting the loop or circuit, as well as on I the permeability of the medium. With a linear medium, self-inductancc does not

depend on the current in the loop or circuit. As a matter of fact, it exists regardless of I whether the loop or circuit is open or closed, or whether it is near another loop or circuit.

A conductor arranged in an appropriate shape (such as a conducting wire wound '

as a coil) to supply a certain amount of self-inductance is called an inductor. Just as t

a capacitor can store electric energy, an inductor can storage magnetic energy, as we 1 shall see in Section 6-12. When we deal with only one loop or coil, therc is no need to carry the subscripts in Eq. (6-1 14) or Eq. (6-1 15), and inductunce without an adjective 1 will bc taken to mean self-inductance. The proccdurc for determining the self-in- f ductance of an inductor is as follows: 5 ' '

1. Choose an appropriate coordinate system for the given geometry.

2. Assume a current I in the conducting wire.

3. Find B from I by Ampire's circuital law, Eq. (6-9), if symmetry exists; if nbt, ! '

Biot-Savart law, Eq. (6-31), must be used. ; 1

\vith orle meability nd hence S

I

(6-1 12)

I not with

(6-113)

1t curreut

((0 4)

(6-115)

e and the file11 as on does not

ardless of :r loop or

re wound Ir. Just as rgy, as we !o need to . ad/\.e le sell-in-

ts; if not,

6-11 / INDUCTANCES AN3 INDUCTORS 235 t s

4. Find the flux linking with each turn, Q, from B by integration,

B = B a d s , I Ss

' where S is the area over which B exists and links with the assumed currenr. 5. Find the flux linkkge A by multiplying B by the number of turns.

6. Find L by taking !he ratio L = All.

Only a slight modification of this proceduk is needed to determine the mutual inductance L I Z b e t d e n two circuits. After choosing an appropriate coordinate system, proceed as fd~lows:.Assume I , + find B, - find B,, by integrating B, over surface S2 -. find fluk linkage A, , = N,(P,, -i find L, , = A, , / I ,

Example 6-12 Assdme I? turns of wire are tightly wound on n toroidal frame of a rectangular cross section with dimensions as shown in Fig. 6-20. Then assuming the permeability of the medium to be ,uo. find the self-inductance of the toroidal coil.

Sulution: It is clear jhat the cylindrical coordinate system is approprlate for t h ~ s problem because the toroili is symmetrical about its axis. A s s u m q a currenr I in the conducting wire, we find. by applying Eq. (6-9) to a circular path wlth radius r (u < r < h):

This result is obtained because bath B, and ra re constant around the circular path C, Since the path encircles a total current NI, we have

Fig. 6-20 A closely wound toroidal ,coil (Example 6-12).


and , .-,, , ,. ,. . . i . . PoNI - B, = -. 2nr

Next we find

The flux linkage A is N@ or poN21h b

A=- In -. . 2n u Finally, we obtain

A poN2h b L=-=- In- (H). I 2 a

(6-116)

We note that the self-inductance is not a function d?ffor a constant nmiium permeability). The qualification that the coil be closely wound on the toroid is to minimize the linkage flux around the iliriividual turns of the wire.

Example 6-13 Find the inductance per unit length of a very long solenoid with air core having n turns per unit length.

Solution: The magnetic flux density inside an infinitely long solenoid has been found in Example 6-3. For current I we have, from Eq. (6-13),

B = ,uonI,

which is constant inside the solenoid. Hence,

where S is the cross-sectional area of the solenoid. The flux linkage per unit length is

Therefore the inductance per unit length is

Equation (6-1 19) is an approximate formula, based on the assumption that the length ol' Ihc solcnoid is vcry n~uch yraltcr th;m thc 1inc;lr dimcnsions of its crow scction. A more accurate derivation for the magnetic flux density and flux linkage per unit length near the ends of a finite solenoid will show that they are less than the values given, respectively, by Eqs. (6-13) and (6-118). Hence, the total inductance of a finite solenoid is somewhat less than the values of L', as given in Eq. (6-1 19), muiti- plicd by the length.

(6-116) n

: meuium :oic' to

d with air

een found

(6-1 17)

. t length is

(6 - 1'1 8)

the 1,- lth m s e ~ - . ~ n . ;e per unit the values tance of a 19), multi-

. q,; .; * :, ,.,.~ , , , r ! - . ;;..:, . .c\; ,, ., ": . ,, -; - : '. . 3 . ::.., .i.i .,. .(: $ :,.; ..,;";-,,: ,, ,!:, .;.: , ..+.:. ,:; > , ,;,, , [,,

I.. ' . i . , . . . I . . ., ..

, ,.., 1 . ,,: i: ' : r ; . . . ' : , f J . , . I

, , '4 . : . , ,

. ! a ; ,! . ~

, . 6-11 ,'I I ~ D U C T A ~ L C S AND INDUCTORS 237 ,

. : , ,

C 4,

The following isia significant obser&ion hbout the results of the previous two examples: The self4 ductance of wire-wound ihductors is proportional to the square ~f the number of tu ps. 1 t

Example 6-14 ~ n ' k i r cdaxial transmission lide has a solid inner conductor of radius a and a very thin outer conductor of inner rahius b. Determine the inductance per unit length of the lide.

!' ' .

Solution: Refer to i g . 6721. Assume that a &rent I flows in the inner conductor and returns via the h e r donductor in the othet direction. Because of the cylindrical symmetry, B has onl) a $-component with diherent expressions 111 ihc two resions: (a) ins~de the inner cbndu;tor, and (b) between the inner and outer conductors. Also '

assume that the currlk!nt I is liniformly distributed over the cross section of the Inner conductor.

a) Inside the inner cbndu :tor, O l r l a .

From Eq. (6-lo],

b) Between the inner and outer conductors,

u ~ r ~ h . From Eq. (6-1 11,

Now consider an:annclar ring in the inner conductor between radii r and r i cir. The current in a unit length of this annular ring is linked by the flux that can be obtained by integrating Eqs. (6-1 20) and (6-121). We have

Fig. 6-21 Two views of a coaxial transmission line (Examplz 6-14).


But the current in the annular ring is only a fr&tion (2nr dr/na2 = 2r dr/a2) of the total current I.' Hence the flux linkage for this annular ring is

2r dr dA' = - a2

dW. (6-123)

The total flux linkage per unit length is , . .

The inductance of a unit length of the coaxial transmission line is therefore

The first term po/8n arises from the flux linkage internal to the solid inner conductor; it is known as the internal inductance per unit length of the inner conductor. The second term comes from the linkage of the flux that exists between the inner and the outer conductors; this term is known as the external inductance per unit length of the coaxial line.

--.

Before we present some examples showing how to detkrmine the mutual indue- . tance between two circuits, we pose the following question about Fig. 6-19 and

Eq. (6-111): Is the flux linkage with loop C 2 caused by a unit current in loop C, equal to the flux linkage with C , caused by a unit current in C,? That is, is it true that

L I Z = L 2 , ? (6-125)

We may vaguely and intuitively expect that the answer is in the affirmative .because of reciprocity." But how do we prove it? We may proceed as follows. Combining Eqs. (6-107), (6-109) and (6-Ill) , we obtain

N2 L I Z = - L2 B, . ds,. I ,

(6-126) ' .I J .

It is assumed that the current is distributed unifohnly in the inner conductor. This assumption does not hold for high-frequency AC currents.

. > I , . . i ' j ;

4 ' n2) of the . I.,!:

I 1 '

. I

(6-123) r r

' i

\nductor; <tor. The .r and the length of

i;il induc- 5-19 and i loop C 1 . is it true

: "because ornbining

n

ion does not

a But, in $ew of Eq. (6,14), B1 can be written as the curl of a vector magnetic potential A,, B1 = V x A,. Welhave

< : : !

N2 ' L12 = - (V x A l ) . ds2 ' f . 11

I

In Eqs. (6-127) and (6-128), the contour integrals are evaluated only once over the periphery of the loogs C, and C , respectively-the effects of multiple turns habinp been taken care of sebarately by the factors N , and N , . Substitution of Eq. (6-128) in Eq. (6-127) yields

where R is the distance between the differential lengths dt', and d&. It 1s customary to write Eq. (6-129aj as

' I whcre N , and N 2 h d ~ c been absorbed in the contour integrals over the circzilt~ C, and C 2 from one end to the other. Equation (6-129b) is the Neumann formela for mutual inductance. U is a general formula requiring the evaluation of a double line integral. For any givdn problem we always first look for symmetry conditions that may simplify the dekrmination of flux linkage and mutual inductance without resorting to Eq. (6-129b) directly.

It is clear from Eq. (6-1296 that mutual inductance is a property of the geometrical shape and tilt physical arrangement of coupled circuits. For a linear medium mutual inductance is proportional to the medium's permeability and is independent of the currents in the circuits. It is obvious that interchanging the subscripts 1 and 2 docs not31ange thd value of ille double integral: hence an slfirmative answer to thc qitcstiot~ poscd in Eq. (0-125) followvs. This is ;In ini1wrt:lnt conclusion bccnuss it allows us to use the simpler of the two ways (flnding L,, or L,,) to determine the mutual indu~tance .~

' In circuit theory books the symbol M is frequently used to denote mutual inductance.

. . . ( ?

I , . ,

, ', rk b , 9 $ ~ d - ; f .

, I . Y 8 r

Fig. 6-22 A solenoid with two windings (Example 6-15). -

Example 6-15 Two coils of turns N , and N2 are wound concentrically on a straight nonmagnetic cylindrical'core of radius a. The windings have lengths C, and C, respectively. Find the mutuil inductance between the coils.

Solution: Figure 6-22 shows such a solenoid with two concentric windings. Assume current I, flows in the inner coil. From Eq. (6-117) we findthat the flux Q I 2 in the solenoid core that links with the outer coil is - -

Since the outer coil has N 2 turns, we have

Hence the mutual inductance is

L12 = - - - 5 NI N Z x a 2 . ( H ) . , (6-130) 11 el

Example 6-16 Determine ihe mutual inductance between a conducting triangular loop and a very long straight wire as shown in Fig. 6-23.

Solution: Let us designate the triangular loop as circuit 1 and the long wire as circuit 2. If we assume a current I , in the triangular loop, it is difficult to find the magnetic flux density B, everywhere. Consequently, it is difficult to determine the mutual inductance L I Z from A, J I , in Eq. (6-1 11). We can, however, apply Am*re's circuital law and readily write the expression for B2 that is caused by a current I , in the long straight wire.

ly on a i d, and

Assume i r h

.A

big. 6-23 A conducting triahgular loop and a long straight wire (Example 6-16).

The flux linkage A,, = @,, is

where d s l = a + z , d r . (6-133)

The relation between z and r is given by the equation of the hypotenuse of the triangle:

l z = - [r - (d + b)] tan 60" = -a[. - (dm+ b)]. (6-134)

Substituting Eqs. (6-1311, (6-133), and (6-134) in Eq. (6-132), we have

- - ~ [ ( d + b ) 1 h ( l + ~ - b ] . 277

Therefore, the mutdd inductance is 6 t (H).

(6-135)

6-12 MAGNETIC ENERGY t

So far we have dishssed self- and m u t ~ a l ihductances in static terms. Because inductances depend on the geometrical shape and the physical arrangement of the conductors constitutlhg the circuit?, and, for a linear medium, are independent of the


currents, we were not concerned with nonsteady currents in the defining of inductances. However, we know that resistanceless inductors appeat as short-circuits to steady (DC) currents; it is obviously necessary that we consider alternating currents when the effects of inductances on circuits and magnetic fields are of interest. A general consideration of time-varying electromagnetic fields (electrodynamics) will be deferred until the next chapter. For now we assume quasi-static conditions, which imply that the currents vary very slowly in time (are low of frequency) and that the dimensions of the circuits are very small compared to the wavelength. These conditions are tantamount to ignoring retardation and radiation effects, as we shall see when electromagnetic waves are discussed in Chapter 8.

In Section 3-1 1 we discussed the fact that work is required to assemble a group of charges and that the work is stored as electric energy. We certainly expect that work also needs to be expended in sending currents into conducting loops and that it will be stored as magnetic en'ergy. Consider a single closed loop with a self- inductance L, in which the current is initially zero. A current generator is Lonnected to the loop. which increases the current i , from zero to I ,. From physics we know that an electromotive forcc (eml) will bc induced in the loop that opposes the current change.+ An amount of work must be done to overcome thisiiitluced emf. Let o, = L, di,/dt be the voltage across the inductance. The work required is

Since L , = @ , / I l for linear media, Eq. (6-136) can be written alternatively in terms of flux linkage as

which is stored as magnetic energy. Now consider two closed loops C 1 and C, carrying currents i, and i,, respectively.

The currents are initially zero and are to be increased to I , and I,, respectively. To find the amount of work required, we first keep i, = 0 and increase i , from zero to I ,. This requires a work W, in loop C,, as given in Eq. (6-136) or (6-137); no work is done in loop C,, since i , = 0. Next we keep i, at 1, and increase i, from zero to I , . Because of mutual coupling, some of the magnetic flux due to i2 will link with loop C,, giving rise to an induced emf that must be overcome by a voltage v, , = L,, di,/dt in order to keep i, constant at its value I , . The work involved is

At the same time, a work W,, must be done in loop C , in order to counteract the induced emf and increase i, to I , .

WZ2 = 4L21$. (6-139)

' The subject of electromagnetic induction will bc diadussed in Chnpter 7.

ics) will a t

,- r, which that the : condi- ' (

hall see \

3 group :cl t h ; ~ t r ~ c l that a self-

nnccted ow hhat c u r 7 7 rt t', -

;I terms

xtlvely. :ely. To 70 to I,. work is 0 to 1 , . 30p c1,

diJdt

i ,;:- ?rj3. .. -The total amount of kork done in raising'the dlrfents in loops C l and C2 from zero to I, and I,, respectitrely, is then the sum of wj Wz,, and W2,.

' C , k2 = f L J : + L ~ J ~ ~ ~ + fL21:

Generalizing this reiplt to a system of N lo&s carrying currents I,, I,, . . . . I,,, we obtain

which is the energy stbretl in the magnetic field. For a current I flowing in a slnglz inductor with inductahce L, the stored magnetic energy is

It is instructive to1 dclivc Eq. (6-141) in an altcrnativc way. Conbider a typlcal " kth loop of N magnetically coupled loops. Let v, and i,, be respectively, the voltage

across and the current in the loop. The work ddne to the kth loop in time tlt 1s

where we have used the relation u, = dlp,/dr. Note that the change. dg,. in the flux +k linking with the kth loop is the result of the changes of the currents in all the coupled loops. The djfferential work done to, or the differential magnetic energy stored in, the system h then

! ' N N

I dWm = 1 dWA = i,d+,. (6 - 143)

k = 1 k = 1

The total stored energy is the integration of rlI.Yv, and is independent of the manner in which the final valttes of the currents and fluxes are reached. Let us assume that all the currcnts and ll,&txcs are brought to thcir final values in concert by an equal fraction cc that increases from 0 to 1; that is, i, = crl,, and 4, = a@, at any instant of time. We obtain the .tdtal tnag~telicr energy:

N


which simplifies to Eq. (6-137) for N = 1, as expected. NO^^ that, for linear media, N

3

- @k = L,kI,,

we obtain Eq. (6-141) immediately.

. F 6-12.1 Magnetic Energy in Terms of Field Quantities

Equation (6-145) can be generalized to determine the magnetic energy of a continuous distribution of current within a volume. A single current-carrying loop can be considered as consisting of a large nimber, N, of contiguous filamentary current elements of closed paths C,, each with a current AI, flowing in an infinitesimal cross-sectional area Aa; and linking with magnetic flux a,.

1 where S, is the surface bounded by C,. Substituting Eq. (6-146) in Eq. (6-145), we have

Now,

As N + a, Av', becomes dv' and the summation in Eq. (6-147) can be written as an integral. We have

I Wm = f 6, A J dv' (J), ( (6 - 148)

where V' is the volume of the loop or the linear medium in which J exists. This volume can be extended to include all space, since the inclusion of a region where J = 0 does not change W,. Equation (6-148) should be compared with the expression for the electric energy We in Eq. (3-140).

It is often desirable to express the magnetic energy in terms of field quantities . B and H instead of current density J and vector potential A. Making use of the vector

identity, V ~ A X H ) = H . ( v x A ) - ' A ~ V x H),

(see Problem P.2-23), we have

A . ( V x H ) = H - ( V x A ) - V . ( A x H ) or

A . J = H . B - V * ( A X H). (6-149)

volume x J = O sion for

4

[:I , ; . , l' 6-12 / MAGNETIC ENERGY 245

; 1 5 ,

substitution ,of ~ h . (4-14$ in Eq. (6-148), we Obtain * 4 4~ = 4 yv, H B du' - n

t 6 . 1 ~ X H) . a. ds'. (6-150) 11 , "

i In Eq. (6-150) we hare applied the divergence theorem, and S. is the surface boundlng V'. If V' is taken to be sufficiently large, the pdints on its surface S' will be very far from the currents. ~b those far-away points, tHe contribution of the surface integral in Eq. (6-150) tends. to zero because Blls bfT as l /R and /HI falls O K as 1/R2, as can be seen from E ~ $ (6-22) and (6-31) respectively. Thus, the rnagnirude of (.4 x H) decreases as 1 / ~ ' , vqhereas at the same time, the surface Sf increases only as R2 When Rapproaches ihfinity. Illcsarf:~;~ i n ~ c p ~ ~ ; ~ l i l l EL,. (6 - 150) Y:II~~SIICS. \lit hii\c 1lic11

Noting that H = B/p, we wn write Eq. (0-151h) in two nlternatiile form,:

and

The expressions in E ~ S . (6- lSla), (6-151 b), and (6-15 lc) for the magnetlc energy W, in a linear medium are analogous to those of electrostatic energy U: in, reipcc- tively, Eqs. (3-146a), (3 -l46b), and (3-146c). a

If we define a m g n e r k energy demiry, w,, such that its volume integral equals the total magnetic energy ,

we can write w,,, in tHtee diiferent forms:

w,,, " F-I . B (J/m3) or

By using Eq. (6-142), w can often determine self-inductance more easily from stored magnetic energy calculated in terms of B and/or H, than from flux linkage.


. .

We have . , . , . I

Example 6-17 By using stored magnetic energy, determine the inductance per unit length of an nir coaxial transmission line that his a solid inner conductor of rtldius e and a very thin outer conductor of inner radius h.

Solution: This is the same problem as that in Example 6-14, where the self-inductance was determined through a consideration of flux linkages. Refer again to Fig. 6-21. Assume a uniform current.1 tlows in the inner conductor and returns in the outerconductor. The magnetic enprgy per unit length stored in the inner conductor is. from Eqs. (6- 120) and (6- 15 1 b),

1 W;, =-J," B;, 2nr,lr

3 1 0 -

-1 - J: r3 dr = -

4na4 '(o" (Jim). . 16n

(6-155a)

The magnetic energy per unit length stored in the region between the inner and outer conductors is, from Eq. (6-121) and (6-151b),

1 Wk2 = - [ ~ ; , 2 n r dr

2'(0

- pol2 b - g [ ; d r = - - I n - (J/m). 4n 4n a

(6 - i 5513)

Therefore, from Eq. (6-154), we have

2 L' = - (Wm, + Wk,)

I2

which is thc same as Eq. (1-124). Tli: pruccdurc used in this solution is compnr;llively simpler.

MAGNETIC FORCES AND TORQUES

In Section 6-1 we noted that a charge q moving with a velocity u in a magnetic field with flux density B experiences a magnetic force F,,, given by Eq. (6-4), which is repeated below:

F, = qu x I3 (N). (6- 156)

, , .( , , . . .

, ... . , , ~. I , . .

' unit

I ~ U C -

) Fig. n the tor is.

r' 155. I

out

dP with a rross-sectional area S. If there are u in the direction df dP,

,'

(6- 157)

where 9 , is the charge bn ea,ch charge carrier. The two expressions in Eq. (6-157) are equivalent since u d)d cl< have the same dirwtion. Now, since NqlSlul equals the current in the condiietor,,,we can write E ~ . (6-*157) as

1.

The magnetic force on a conlplete (closed) circuit of contour C that carrles a current '

I in a magnetic field B ill then

(6- 159) 1 . - - - - - - - - .

When we have two circuits carrying currents Ir and I , respectively, the situatmn is that of one current-cafryingcircuit in the magnetic field of the other. In the presence of the magnetic field Bll, which was caused by the current I, in C,, the force F,, on circuit C, can be written +

I I

where B,, is, from the hiot-bavart law in Eq. (6-31),

Combining Eqs. (6-1608) and (6- 160b), we obtain

which is AmpPre'slaw o j force between two current-carrying circuits. It is an inverse- square relationship and should be compared with Coulomb's law of force in Eq. (3- 17) between twaaationary charges, the latter being much the simpler.

The force ~y;on circuit C,, in the presence of the magnetic field set ua br the . . current I , in circuit C,, can be written from Eq. (6-16fa) by interchanging the subp ip t s 1 and 2.

, 1


I . . . I , However, since dt', x (dl', p a.,,) # -dCl x.(d& x a.,,), we inquire whether this means F z l # -F,,- that is, whcther Nowton's third law governing thc l'orcrs of action and reaction fails here. Let us expand the vector triple product in the integrand of Eq. (6-161a) by the back-cab rule, Eq. (2-20).

Now the double clbsed line integral of the first term on the right side bf E ~ . (6-162) is

In Eq. (6-163) we havemade use ofEq. (2-81) and the relatio3q.(l/R2,)= - aR2,/RZl. The closed line integral (with identical upper and lower limits) of d(l/R,,) around circuit C1 vanishes. Substituting Eq. (6462) in Eq. (6-161a) and using Eq. (6-163), we get

Po FZ1 = -- ' dP2) 471 l 2 c c R:,

(6- 164)

which obviously equals - F,,, inasmuch as a,,, = -aR2,. It follows that Newton's ihird law holds here, as expected.

Example 6-18 Determine the force per unit length between two infinitely long parallel conducting wires carrying currents I, and I , in the same direction. The wires are separated by a distance d.

Solution: Let the wires lie-in the yz-plane and be parallel to the z-axis, as shown in Fig. 6-24. This problem is a straightforward application of Eq. (6-160a). Using F;, to denote the force per unit length on wire 2, we have

.- F;2 =I,(% x BIZ), (6-165) where B12, the magnetic flux density at wire 2. set up by the current I, in wire 1, is constant over wire 2. Because the wires are assumed to be infinitely long and cylindrical symmetry exists, it is not necessary to use Eq. (6-160b) for the determination

,,. . of BIZ. We apply Amptre's circuital law, and write, from Eq. (6-ll),

4- around 6-I/

:ly lpng 2e wires

i . .2"

, . , . 6-13 1 %GN~TIC FORCES AND TORQUES 249 . . . ,:.;; !

x ' ! A 3 .. , Fig. 6-24 Force bitween two parallel I current-carrying wires (Example 6-18).

Substitution of Eq. (6-16t) in Eq. (6-165) yields I

We see that the force on w~re 2 pulls it toward wire 1. Hence, the force between two . wires carrying currents iu the some direction is one of attractioll (unl~ke the force

between two charges df the same polarity, which is one of repulsion). It is trivial to prove that F;, = - fl: i 2 - - a,(p,1~1,/2nd) and that th.e force between two wires carrying currents in opeosite directions is one of repulsion.

f

Let us now considir a small circular loop of radius b and carrying a current I in a uniform magnetic field, of flux density B. It is convenient to resolve B into two components, B = B, + Blitwhere B, and Bll are,,respectively, perpendicular and parallel to the plane of the loo& As illustrated in Fig. 6-25a, the perpendicular component B, tends to expand the lobp (or contract it, if the direction of I is reversed), but B, exerts nc net force to move tlie loop. The parallel component Bll produces an upward force

(a) (W

Fig. 6-25 A circular loop in a uniform magnetic field B = B, + BIl.

250 STATIC MAGNETIC FIELDS 1 6 i ' I

1 1 1 dF, (out from the paper) on element dt, and a downward force (into the paper) dF2 = -dFl on the symmetrically located element d t 2 , as shown in Fig. 6-25b. I

> Although the net force on the entire loop caused by BII is also zero, a torque exists I

that tends to rotate the loop about the x-axis in such a way as to ulign the magnetic I

field (due to I) with the external BII field. The differential torque produced by dF, and dF, is I !

dT = ax(dF) 2b sin 4 = a,(I dt BII sin 4j2b sin 4 = a,21b2BII sin2 4 d4, (6-168)

If the definition of thc rnagnctic dipolc momcnt in Eq. (6--46)-is~!;cd,

where a, is a unit vector in the direction of the right thumb (normal to the plane of the loop) as the fingers of the right hand follow the direction of the current, we can write Eq. (6-169) as

The vector B (instead of B,,) is used in Eq. (6-170) because rn x (BL + BII) = rn x Bll . This is thc torque that aligns thc microscopic mclgnclic dipolcs in magnctic matcrials and causes the materials to be magnetized by an applied magnetic field. It should be remembered that Eq. (6-170) does not hold if B is not uniform over the current- carrying loop.

Example 6-19 A rectangular loop in the xy-plane with sides b, and b, carrying a current 1 lies in a unijorm magnetic field B = a,B, + ayBy + a$,. Determine the force and torque on the loop.

Solution: Resolving B into perpendicular and parallel components B, and B,,, we have

BL = a$, (6-171a)

BII = a,B, + ayBy. (6-171b)

Assuming the current flows in a clockwise direction, as shown in Fig. 6-26, we find that the perpendicular component a$, results in forces Ib,B, on sides (1) and (3) and

8

: paper) I I

6-25b. - , i ' IC exists , . , lagnetic a

by dF, ' ( ' i I

m x Bil. naterials lould be current-

.rrying a the force

. we find d (3) and -

Fig. 6-26 A rectangular loop in a ,, uniform ~nagnetlc field (Example 6-19).

i

forces Ib,B, on sides (2) and (41, all directed toward the center of the loop. The vector sum of these four contracting forces is zero, and no torque IS produced.

The parallel combonsnt of the magnetic flux denslty. Bl l , produces the following ' forces on the four s idb:

Again, the net force bn the loop, F, + F, + F,'+ F,, is zero. However, these forces result in a net torque' that can be computed as follows. The torque TI, , due to forces '

F, and F, on sides (1) and (3), is

TI, = axIblb2B,; (6- 173a)

the torque T2,, due to forces F, and F, on siddk ( 2 ) and (4), is

T2, = -a,lb,blBx. , The total torque on the rectangular loop is, the4

1

Since the magnetic(domenl of the loop is rn = - aJb,b,, the result in Eq. (6- 174) isexactly T = rn % (a$x -j- aYBJ = rn x I% en&, in spite of the fact that Eq. (6- 170) was derived for a iircular loop, the torque f o d u l a holds for a p!anar loop of any shape as long as it is lbcated in a uniform magnetic field.

, . 6-13.1 Forces and Torques in Terms of . . a .. Stored Magnetic Energy

All current-carrying conductors and circuits experience magnetic forces when situated in a magnetic field. They are held in place only ifmechanical forces, equal and opposite to the magnetic forces, exist. Except for special symmetrical cases (such as the case of the two infinitely long, current-carrying, parallel conducting wires in Example 6-18), determining the magnetic forces between current-carrying circuits by Ampere's law of force is a tedious task. We now examine an alternative method of finding magnetic forces and torques based on the principle of virtual displacement. This principle was used in Section 3-1 1.2 to determine electrostatic forces between charged conductors. Here consider two cases: first, a system of circuits with constant magnetic flux linkages; and, second, a system of circuits with constant currents.

, System of Circuits with Constant Flux Linkages If we assume that no changes in flux linkages result from a virtual differential displacement d f of one of the current- carrying circuits, there will be no induced emf's and the sources will supply no energy to the system. The mechanical work, F, df . done by the system& at the expense of a decrease in the stored magnetic energy, where F, denotes the force under the constant flux condition. Thus,

F , . d f = -dW,= -(VWm).df,

from which we obtain

pzGk7q In Cartesian coordinates, the component forccs arc

If the circuit is constrained to rotate about an axis, say the z-axis, the mechanial . work done by the system will be (T,), dm and

(TaL = - - (6-178) X . ,

. which is the z-component of the torque acting on the circuit under the condition of constant flux linkages.

indirl$ *. Thid '

I

large8 t ,

~gnetik!

anical

n

dition b

Fig. 6-27 An eldttromagnet (Example 6-20). !

Example 6-20 Codbider the electromagnet in Fig 6-27 where a current I in an N-turn coil producck a l lux O in thc magnetic circuit. Thc cross-sectional area of thc core is S. Dctcrdinc 1.11~ lifting forcc on the armature.

Solution: Let the armature take a virtual diyplacement ily (a differential increase . in y) and the source.be adjusted to keep the llux (b constant. A displaccment of the

armature changes oqly the length of the air aps; consequently, the disphcement changes only the magnetic energy stored in t e two air paps. We have. from Eq, (6-151b), 1 b

An increase in the length (a positive dy) increases the stored magnetic energy if (r, is constant. ~ s i t l i Eq. ,(6-177b), we obtain

Here, the negative sign indicates that the force tends to reduce the air-gap length: that is, it is a force 01 attraction. -.

I'

--I- Systcm of Circuits wlL Constant Currcnts In this case the circuits arc connected to current sources that :coullter:ict the induced emf's resulting from changes in flux linklgcs that are caUsed by a virtual displaccment dP. The work done or energy supplied by the sour& is (see Eq. 6-144),

-

254 STATIC MAGNETIC FIELDS / 6 . .

This energy must be equal to the sum of the mechanical work done by the system dW (dW = F, -dt, where F, denotes the force on the displaced circuit under the

- constant current condition) and the increase in the stored magnetic energy, dW,,. That is,

dW, = dW + dWm.

From Eq. (6-145), we have

Equations (6-182) and (6-183) combine to give

dW.= F , . d e = dWm , = (vw,) dt

or

which differs from the expression for F, i.n Eq. (6-176).only by a sign change. If the circuit is constrained to rotate about the z-axis, the z-component of the torque acting on the circuit is

The difference between the expression above and (T,), in Eq. (6-178) is, again, only in the sign. It must be understood that, despite the difference in the sign, Eqs. (6-176) and (6-178) should yield the same answers to a given problem as do Eqs. (6-184) and (6-185) respectively. The formulations using the method of virtual displaccment unclcr conscant llux linkagc :mcl conhtanl currcrlt col~ditions arc simply two means of solving the same problem,

Let us solve the electromagnet problem in Example 6-20 assuming a virtual displacement under 'the constant-current condition. For this purpose, we express Wm in terms of the current I :

W, = ~ L I ~ , (6-186)

where L is the self-inductance of the coil. The flux, @, in the electromagnet is obtained by dividing the applied magnetomotive force (NI) by the sum of the reluctance of the core (9,) and that of the two air gaps (2y/p,S). Thus,

. I f the tori

6-185)

n. only 6-176) 6-184) cement means

, ,

virtual rcss Wm

1 " ~nduct&e L ib'kqtt~1 tdkux linkage per ynidcurrent.

1 2 . ; r: N@ ' s ' ,N2

\ 4 ; L=-= "! I <W'

'Combining Eqs. (63184) and (6-186) and usidg Eq. (6-188), we obtain . 4

I 2 dL N I 2

, F r = a y i - - --a,,"( ) 2 d y CLoS WC + 2ylCLoS '; - a2' { - -ay+ (N), (6-189) I

clos

which is exactly the bame as the F, in Eq. (6-180). .

6-13.2 Forces and Torques in Terms of Mutual Inductance I j -

II The method of virtual displacement for cohstant currents provides a powerful technique for detettnining the forces and torques between r ~ g d current-carrylnp circuits. For two circuit> with currents I, and I , , self-inductances L, and L2, and mutual inductance t,,, the magnetic energy is, from Eq. (6-140),

I: W, = +L,I: + ' L , , I ~ I , + $L21:. (6-190)

If one of the circuiti is given a virtual displacement under the condition of constant currents, there wodd be a change in Wm and E q (6-184) applies. Substirution of Eq. (6-190) in Eq. (6-184) yields

0.1 Similarly, we obtain from Eq. (6-185),

Example 6-21 Determine the Torce between two coaxial circular coils of radii b, m d 6, ~ c p i i r i ~ t ~ l hy iI ~iiisi~~nrn 11 which is m ~ c h Iiwgcr ~ I ~ : I I I tllc radii ( t i >> b,, h,). , I , I IC coils &:;)llsisl of N ill14 N 2 c10scIy woudd turns and carry currents I , and 1,

rcspcctivcly,

Solution: This problem is rather a difficult one if we try to solve it with Ampere's law of force, as exdressed in Eq. (6-1614. Therefore we will base our solution on Eq. (6-191). First, We determine the mutual inductance between the two coils. In

Fig. 6-28 Coaxial current-carrying circular loops (Example 6-21). -

I-.-

Example 6-7 we found, in Eq. (6-43), thavector potential at a distant point, which was caused by a single-turn circular loop carrying a current I. Referring to Fig. 6-28 for this problem, at the point P on coil 2 we have A,, due to current I , in coil 1 with N , turns as follows:

PoNiI1b: 4412 = a, -----_ sin 8 4R2

In Eq. (6-193), z, instead of d, is used because we anticipate a virtual displacement, and : is to be kept as a variable for the time being. Using Eq. (6-193) in Eq. (6-24), we find the mutual flux.

The mutual inductance is then, from Eq. (6-Ill),

I .

i' REVIEW QUESTIONS 257

' i I :

On coil 2, the fotc+dud to the magnetic field of bil 1 can now be obtained directly by substituting Ed: 16-195) in Eq. (6-191). $ '

which can be written ad

The ncgative sign in Eq, (6-196) indicates that I;,, is a forcc of attraction Tor currents flowing in the same directioa. This force diminishes very rapidly as the inverse fourth power of the distance of separation.

REVIEW QUESTIONS

R 6 - 1 What is ths.expression for the force on a test charge y that moves wrth vrloc~ty u in a magnetic field of flux density B?

H.6-2 Vsrify that tesla (T), the unit for magnetrc flux denslty, is the same as volt-xcond per square meter (V.s/niZ).

R.6-3 Wrrte Lorentz's force equation.

R.6-4 Which postulate bf magnetostatics denies the existence of isolated magnetic charges?

R.6-5 State the law of cthservation of magnetic flux.

R.6-6 State Ampere's chcuital law.

R.6-7 In applying Anlpeh's circuital law nlust the path of integration be circular? Explain.

R.6-8 Why cannot the B-field.of ad infinitely long straight, current-wrrying conductor have n componolt in thc dircctlon or the currcnl?

R.6-9 DO the for iulas for R. As derr;h in Eqr (6-lp) and (6-11) for .L round conductor, apply 10 c o ~ ~ d u c t o r Lavillg i l UllliM crobh SIX~IOII of the same ilrea rild carryrng the ramr currcnt? ~ x ~ l n i 2 ' - -

H.6-10 in what niasdcr does tllc U-lield of an rnlinitely long stnight filament urrying a drren current I vary with distance?

R6-11 Can B-field exist b a p o d conductor? Elplain.

R.6-12 Define in words udrtor magnetic potential A . What is its S1 unit?


\

R6-13 What is the relation between magnetic flux density B and vector magnetic potential A? Give an example of a situation where B is zero and A is not.

R6-14 What is the relation between vector magnetic potential A and the magnetic flux through a given area?

R.6-15 State Biot-Savart's law.

R.6-16 Compare the usefulness of Arnpkre's circuital law and Biot-Savart's law in determining B of a current-carrying circuit.

R6-17 What is a magnetic dipole? Define magnetic dipole moment.

R6-18 Define scalar magnetic potential Vm. What is its SI unit?

R6-19 Discuss the relative merits of using the vector and scalar magnetic potentials in magnetostatics. r

R6-20 Define magnetization vector. What is its SI unit?

R.6-21 What is meant by "equivalent magnetization current densities"? What are the SI units for V x M and M x a,?

-1. R.6-22 Define magneticfield intensity vector. What is its SI unit?

R6-23 Define magnetic susceptibility and relati& permeability. What are their SI units?

R.6-24 Does the magnetic field intensity due to a current distribution depend on the properties of the medium? Does the magnetic flux density?

R.6-25 Define magnetomotive force. What is its SI unit?

R6-26 What is the reluctance of a piece of magnetic material of permeability p, length C, and a constant cross section S? What is its SI unit?

R.6-27 An air gap is cut in a ferromagnetic toroidal core. The core is excited with an mmf of N1 ampere-turns. Is the magnetic field intensity in the air gap higher or lower than that in the core?

R.6-28 Define diamagnetic, paramagnetic, and ferromaynetic materials.

R6-29 What is a magnetic domain?

R6-30 Define remanent j4ux density and coercive field intensity.

R6-31 Discuss the difference between soft and hard ferromagnetic materials.

. R6-32 What is curie temperature?

R6-33 What are the characteristics of ferrites?

R.6-34 What are the'boundary conditions for magnetostatic fields at an interface between two $jp , yi ,. different magnetic media?

R.6-35 Explain why magnetic flux lines leave perpendicularly the surface of a ferromagnetic medium.

I I -

R.. at th.

R.I slr

R.c tur

R.I thl

R.c

R. I in

- R.r St [

A. ( cu: CO'

PROBLEV \

P.6 reg ch.

P.6 f i c ~

Dl

/1 P. t.

4 the f ' O U - pic

P.C an to

' ,.,. . , . . , . ,.:. 1. , , ',':,.: :

potential : :.;;j,jj::.;;1, ' , ; ,.,$ ; ', :.

. . . . . ' , , , '" , . '; ' ,\ .. :. :1;4j:: {,-)$ t ~ l ~ o u y h ,;..!; 2 : ' 6 <: L~ ! ;

. , 'j . i i.? ! ;;. 1 :: !! , . ,',', ; . ,> . . .

, , , .?;'',!. ' I . 8 , , , .

: S1 units

t h P. and

n mmf of nat in the

magnetic

PROBLEMS 259

' \ ,; 4 ! 1,

R6-36 What boundaryi condi!ion must the tangenVal components of magnetization satisfy , at an intcrfacc? If regiorl2 i6 n:onmaynctic, what is the relation betwcen thc surface current and

thc tanycntial compbac~t of M I ? .

R.6-37 Define (a)'the mhtual j?ductance betweka two circuits, and (b) the self-inductance or a single coil. , . . $ , % :

R.6-38 Explain how the,.self-hd~lctance of a wire-wd~nd inductor depends on ~ t s number of turns. b, i , I

R.6-39 In Example 6-f$, would the answer be ;he dame if the outer conductor is not "very thin"? Explain.

1 i

4 R.6-40 Give an expression o i magnetic energy In terms of B dndlor H.

I

R.6-41 Clve thc integral exprebslon for the force on a closed clrcult t h d c m x s d current 1 In a magnetic field B.

R.6-42 Discuss first the net force and then the net torque acting on a current-carrying cllcult situated in a uniform mlgnetq field.

R6-43 What IS the reladon tztween the force and the stored magnetic energy In system of current-carrymg c~rcuits dnder the condition of constant flux linkages? Under the condmon of constant currents?

1

PROBLEMS

P.6-1 A positive point charge q of mass m 1s Injected with a velocity u, = a,u, into the J > 0 region where a uniform magnetic fleld B = a,B, exists. Obtarn the equat~on of motion of the charge, and describe the pdth that the chargc follows.

I

P.6-2 An electron is injected with a velocity u, = a,ilt, into a region where both an electrlc field E and a magnetic field B exist. Describe the motion of the electron if

a) E = a,E, and B = a$,, b) E = -a,E, and B = -a,B,.

Discuss the effect of the relative n~agnitudes of E, and B, on the electron paths of (a) and (b). I

P.6-3 A current I Ilows i t ) (lie inncr I A ~ I W I O I - ( , ~ : ! I I idillrlcly long co:i\i.~I Iinc atld roturnb wa lhc 0 \ 1 1 ~ ~ L X ~ I I ~ U ~ I ~ . , 'I'hc I & ~ I I I S oI' L I I C IIIIICT CL)~~LIC'IOT is 11, and the Illlicr and outer radil of the O U ~ C I ' L ' O I ~ ~ I C ~ W L& 1) I I I \ ~ e ~ ~ c h ~ ~ c l ~ ~ c l y , I ; ~ I I ~ 111e t ~ u ~ g l i c k Ihu clcudy 11 Vor ,dl ucgionh ;mu plot IBI versus r. \

P.6-4 Determine the magnetic flux denslty at a point on the axis of a solenold with radius h and length L, and with a current 1 in its N turns of closely Wound coil. Show that the result reduces to that given in Eq. (6-13) when t approaches infinity.

260 STATIC MAGNETIC FIELDS 1 6 . .

, * , ,

P-6-5 Starting from the e ~ p r e s ~ i o n for vector magnetic potential A in Eq. (6-22), prove that

Furthermore, prove that V . B = 0.

J P.6-6 Two identical coaxial coils, each of N turns and radius b, are separated by a distance d, as depicted in Fig. 6-29. A current I flows in each coil in the same direction.

a) Find the magnetic flux density B = axBx at a point midway between the coils. b) Show that dB,.dx vanishes at the midpoint. c) Find thc relation between h and ( 1 such that r12B,/rl.~2 also vanishes at the rnidnoint .

r - - - -

Such a pair ufcuils arc used to obtain an appruxi~l~ately uniform magnetic field in the midpoint region. They are known as Heli~rholt~ coils.

d (Problems P.6-6).

dP.6-7 A thin conducting wire is bent into the shape of a regular polygon of N sides. A currcnt I flows in the wire. Show that the magnetic flux density at the center is

!JON[ n . B = a , - tan -, 2nb N

where b is the radius of the circle circumscribing the polygon and a, is a unit vector normal to the plane of the polygon. Show also that as N becomes very large this result reduces to that given in Eq. (6-38) with z = 0.

P.6-8 Find the total magnetic flux through a circular toroid with a rectangular cross section of height h. The inner and outer radii of the toroid are u and b respectively. A current I flows in N turns of closely wound wire around thc toroid. Determine the percentage of error if the flux is found by multiplying the cross-sectional area by the flux density at the mean radius.

P.6-9 in certain cxpcriments it is dcsirshle to havc a rcgion of constant magnetic [lux density. This can be created in an off-center cylindrical cavity that is cut in a very long cylindrical conductor

I ' we that . . ' /

, ! - 1 (6-197) , , , : '

tmce (i,

1 current

s to that il

A density. I

.onductor

!,,

carrying a uniform current density. Refer to the cross section in Fig 6-30. The uniform i~rial currcnt density is .I :- ;I,J: Find [hc in;~yi l i l~~dc xnd dirCcii011 of U ill ihc vyiindiicll cavity who* axis is displaced from that of the conducting part by a distance d. (Hint: Use principle of superposition and consider B in thecavity as that due to t d o long cylindrical conducton with radii b and a and current densities J and -J respectively.)

- ,

P.6-10 Prove the following:

t p.6-12 Startrng from the expressron of A In Eq, (6-34) for the vector magnet~c potent~al at a point in the brsecting ~ l a f l e of 8 straight wlre of length 2L that carrles a current 1

a) Find A at pomt k x , y.0) in the blsectrng plane of two parallel r ~ r e s each of length 21, located at y = &!/2 and carrylng equal and obpos~te currents, as shorn in Fig. 6-31.

b) Flnd A due to ecjbal and opposite currents m n very long two-w~re transm~ss~on line. c) Find B from A in part (b), and check your answer against the result obtarned by ~pplying

Ampkre's circuital law.

Fig. 6-31 parallel wires carrying equal and apposite currents (Problem P.6-12).


P.6-13 For the small rectangular loop with sides a and b that carries a current I, shown in Fig. 6-32:

a) Find the vector magnetic potential A at a distance point, P(.u, y,d. Show that it can be put in the form of Eq. (6-45).

b) Determine the magnetic flux density B from A, and show that it is the same as that given in Eq. (6-48).

--- -- P.6-14 For a vector field F with continuous.first derivatives, prove that

& ( v x ~ ) d u = - $ , F ~ L ,

where S is the surface enclosing the volume V. (Hint: Apply the divergence theorem to (A x C ) , where C is a constant vector.)

P.6-15 A circular rod of magnetic material with permeability p is inserted coaxially in the long solenoid of Fig. 6-4. The radius of the rod, a, is less than the inner radius, b, of the solenoid. The solenoid's winding has n turns per unit length and carries a current I.

a) Find the values of B, H, and M inside the solenoid for r < a and for a < r < b. b) What are the equivalent magnetization current densities J, and J,, for the magnetized

rod'?

P.6-16 The scalar magnetic potential, Vm, due to a current loop can be obtained by first dividing thc loop area into many small'loops and then summing up thc contribution of thcse small loops (magnetic dipoles); that is,

where dm = a,I ds. (6-198b)

Prove, by substituting Eq. (6-198b) in Eq. (6-198a), that

where R is the solid angle subtended by the lbop surface at the field point P (see Fig. 6-33).

) (A x C),

i the long no~cl. The

dividing :;lit loops

i

i I \

PROBLEMS

Fig. 6-33 Subdivided current loop for determination of scalaf magnetic potential (Problcm P.6-16).

P.6-17 DO the following by ilsing Eq. (6-199):

a) Determine the scalar magnetic potential at a point on the axis of a circular loop having radius h and carryinb' a current I.

b) Obtain the maghetic flux density B from -Po OK,, and compare the rcsult with Eg. (6-38).

P.6-18 A ferromagnetikpphere of radius b is magnetized uniformly wlth a magnetlation M =

azM0. a) Determrne the equivalent magnetlzatlon current densltles J, m d J, b) Dctcrmlnc the nlagnclic llux densrty at the center of the qhere.

P.6-19 A toroidal lron core of relative permeab~l~ty 3000 has a mean radlus R = 80 (mm) m d .i c~rcular cross section wlth rrd~i ls b = 25 (mm). An .ur gJp fU = 3 (I,,) exlatr, J I I ~ .I cui rc11, i ilou \

in a 5W-turn wlndlng to poducea lnagnetlc flux of (Wb). (See Rg. 6-34.) Neglecting leakage

Pig. 6-34 A toroidal iron core with air gap (Problem P.6-19).


and using mean path length, find

a) the reluctances of the air gap and of the iron core. b) B, and H, in the air gap, and B, and H, in the iron core. C) the required current I .

P.6-20 Consider the magnetic circuit in Fig. 6-35. A current of 3 (A) flows through 200 turns

of wire on the center leg. Assuming the core to have a constant cross-sectional area of (m2) and a relative permeability of 5000:

a) Determ~ne thc magnctlc llux in cach leg. b) Dctcrrninc thc magnetic liclcl intensity i l l c;~ch Icg of l l i c core i111tl i l l the itlr gi~p.

Fig, 6-35 A magnetic circuit with air gap (P~oblem P.6-XI),

P.6-21 Consider an infinitely long solenoid with n turns per unit length around a ferromagnetic core of cross-sectional area S. When a current is sent through the coil to create a magnetic field, a voltage v , = - 1 1 dO/dt is inducod pcr unit length, which opposes the current change. Power P, = - v , I per unit length must be supplied to overcome this induced voltage in order to In- crease the current to I.

a) Prove that the work per unit volume required to produce a final magnetic flux density B j is

W, = JOE' H dB. (6-200)

b) Assuming the current is changed in a periodic manner such that B is reduced from BJ to -B, and then is increased again to BJ, prove that the work done per unit volume for such a cycle of chayge in the ferromagnetic core is represented by the area of the hysteresis loop of the core material.

P.6-22 Prove that the relation V x H = J leads to Eq. (6-99) at an interface between two media.

P.6-23 What boundary conditions must the scalar magnetic potential Vm satisfy at an interface between two different magnetic media?

P.6-24 Consider a plane boundary (y = 0) between air (region 1, p,, = 1) and iron (region 2, P a = 5000).

a) Assuming B, = a,0.5 - aJ0 (mT), find B, and the angle that B, makes with the interface. b) Assuming B, = a,10 + a,OS (mT), find B, and the angle that B, makes with the normal

to the interface.

P.6-25 The method of images can also be applied to certain magnetostatic problems. Consider a straight thin conductor in air parallel to and at a distance d above the plane interface of a magnetic material of relative permeability p,. A current I flows in the conductor.

I . , 6 ~

1 turns ,

/-'

~yllctlc IC licid. Power to 111-

density

,6-200)

rom L?, wlume I of the

I media.

nterface

! ~terfx. normal

h s i d e r :riace of

and these cuhents ~ i r c equidistant from the interface and situated in alr, ii) the magnet~dfield below the boundary plane is calculated from I and - I , , both at

the same lodtion! These currents a r i sitbated in an infinite magneuc material of relative perx&bility p,. <

b) For a long m d b c t o r carrying a current I ahd for p? >> 1, determine the magnetic flux density B at thejoint P in Fig. 6-36.

- ' P k Y )

I @ t I

d C 0 Fig. 6-36 A current-carrying conductor Ferromagnetic rncd~pm ( h 2 - 1 ) , near a ferromagnetlc medium (Problem

P 6-25).

I'.6-26 Dctcrmme the 811-inductance of a toroidrl coil of N turns of wlrc wound on an i l r frame with mcan radius 'ro and a circular cross section of radius b Obtain an rpproxlmate expression assuming b << r,,

P.6-27 Refer to ~ x a r n & 6-13. Dctcrminc lhc inductmcc pcr unit icngtii of tile sir coaxial transmission line anumibg that its outer conqu&r is dot very thin but is of a thickness d.

V P.6-28 Calculate the int$tnal and external inductances per unit length ofa two-wire transmission line consisting of two lohg parallel conducting wires of radius n that carry currents in opposite directions. The wires are separated by an axis- tohis distance d, which is much larger than a.

"P.6-29 Determine the inductance between a very long straight wire and a conducting equilateral triangular ~ o d b , as shown

Fig, 6-37 A long straight- wire and a conducting equilateral triahgular loop (Pr'oblem P.6-29).


Fig. 6-38 A long straight wire and a conducting circular loop (Problem P.6-30).

'

P.6-30 Determine the mutual inductance between a very long straight wire and a conducting circulx loop, as shown in Fig. G -.is.

. P.6-31 Find the mutual inductance between two coplanar rec tangula r -bo~wi th parallel sides, ,-.

as shown in Fig. 6-39. Assume that 11, >> / i , ( l ~ , > w2 > ti).

Fig. 6-39 Two coplanar rectangular loops, h , >> h2 (Problem P.6-31).

P.6-32 Consider two mupled circuits, having self-inductances L, and L,, that carry currents . I , and I , respectively. The mutual inductance between the circuits is M. -.

a) Using Eq. (6-140), find the ratio 1 , / 1 2 that makes the stored magnetic energy W2 a mirimum.

b) Show that M 5 a. P.6-33 Calculate the force per unit length on each of three equidistant, infinitely long parallel wires 0.15 (m) apart, each carrying a current of 25 (A) in the same direction. Specify the direction of the force.

\i P.6-34 The cross section of a long thin metal strip and a parallel wire is shown in Fig. 6-40. Equal and opposite currents I flow in the conductors. Find the force per unit length on the conductors.

currents

:gy W, a

n , pF' "

J I ~ c ~ . _.

ig. 6-40. .h on the

. . I . .I " ' 4

, < PROBLEMS 267

1- - 1 palkillel strip and wire ' codduct& (Problem P.6-34).

\/ P.6-35 Refer to robl lei 6-30 and Fig. 6-38. ~ i h d the force on the crrcular loop that IS exerted by the magnetic field du& to an upward current I, in the long straight wire. The circular loop carries a current I, in th~'counterc1ockwise direction.

"6-36 Assuming the circulal loop in Problem P.6-35 a rotated about its horlaontal axis b j an angle a, find the torque exerted on the circular :oop.

P.6-37 A small circular !urn of wire of radius i , that carrier a steady current I , is @aced at the center of a much larger turn:of wire of radius r, ( r , s r , ) that carries a steady current 1: in the same direction. The angle between the normals of the two circuits is 0 and the small circular wire is free to turn about i ts diameter. Determine the rni~gnitude anci the ~iirnc~ioli of ihc torquc ( 1 1 1 I ~ I C SIT~:III sirct~I:~r wire,

. . P.6-38 A magnetized coinpass needle will line up with the earth's magnetis licld. A srnilil bar magnet (a magnetic dipole) with s magnetic moment I (A.m2) is placed st a disrmce 0.15 (m) from the center of a cornp& needle. Assuming the earth's magnetic flux density at the needle to be 0.1 (mT). find the maximum angle at which the bar magnetecan cause the needle to dcuiatc from the north-south direction How should the bar magnet be oriented?

P.6-39 The total mean length of the llux path in iron for the electromagnet in Fig. 6-27 is 3 (m) and the yoke-bar contaci areas measure 0.01 (m2). Assuming the permeability of iron to be 40%0 and each of the air gaps to be 2 (mm), calculate the mmf needed to lift a total mass of 100 (kg).

- x Fig. 6-41 A long solenoid with iron core pattially drawn (Problem P.6-4) .

p.6-40 A current I flowd in a long solenoid with 11 closeiy wound coll-turns per unlt length. The cross-sectional area of its iron core, which has permeability n S. Detemme the force acting on t h e W . i t is withdrawn to the pos~tion shown in Fig. 6-41.

7 / Time-Varying Fieids and . Maxwell's Equations

7-1 INTRODUCTION

In constructing the electrostatic model, we defined an electric , . .- E. and an electric flux density (electric displacement) vector, C .

. governing dilrcrcntial cclu;~t ions arc

For linear and isotropic (not necessarily homogeneous) media, E and D are related by the constitutive relation

D = E E . (3 -97)

. For the magnetostatic model, we defined a magnetic flux density vector, B, and a magnetic field intensity vector, H. The fundamental governing differential equations are

V . B = O (6-6)

V x l l = J . (6 -68)

The constitutive relation for B and H in linear and isotropic media is

These fundamental relations are summarized in Table 7-1. We observe that, in the static (non-time-varying) case, electric field vectors E and

D and magnetic field vectors B and H form separate and independent pairs. In other words, E and D in the electrostatic model are not ;elated to B and H in the magnetostatic model. In a conducting medium. static electric and magnetic fields may both

, , exist and form an electromagnetostaric$eld (see the statement following Example 5-3 on p. 187). A static electric field in a conducting medium causes a steady current to flow that, in turn, gives rise to a static magnetic field. However, the electric field can be completely determined from the static electric charges or potential distributions.

Table 7-1 hndamental Relations for J$ectr~~tatic and

Governing equations v ~ t ) = ~ V x H = J

~onstitudve Relations (linear a(d isotropic media)

* 1

c related

(3 -97)

r, B, and :quatiom

x s wd I n , -r magnPt+

< nay L-.l , mple 5-3 ' ,

:urrent to : fieldxan ' ributions.

The magnctic field is:a cmscqucnce; it docs not enter into the calculation of the electric field.

In this chuptcr wc will skc that :I ch;inging mlignctic licld gives rise to an dcctric licld. and vice versa.',To explain clcctromagnctic phenomena ~ ~ n d c r time-varying conditions, it is necessary loconstruct an electromagnetic model in which the electric field vectors E :ind 6 arc properly related to the magnetic field vectors B and 'H. The two pairs cf the governing equations in Table 7-1 must thcreforc bc modificd to show a mutual depeqence between the electric and magnetic field vectors in the time-varying case.

Wc will begin witti a r~~ntlan~cnlal postulate that modifics ~ h c V x E cquation in Table 7-1 and leads to Faraday's law of electromagnetic induction. The concepts c;f

transformer emf and rflotional emf will be discussed. With the new postulate we will also need to modify the V x H equatiop in order to make the governing equations consistent with the eqitation of contindty (law uf conservation of charge). The two modified curl equatiohs together with the two divergence equations in Table 7-1 are known as Maxwell's equations and form the foundation of electromagnetic theory. The governing eq~~at ions for electrostatics and magnetostatics are special forms of Maxwell's equations when all quantities are independent of time. Maxwell's equations can be conibined to yield wave equl-itions that predict the existence of electromagnetic waves propagating wit11 the velocity of light. The solutions of the wave equations, especikilly for time-harmonic fields, will be discussed in this chapter.

'7-2 FARADAY'S LAW OF ELECTROMAGNETIC INDUCTION ,

-\ \.

, A major advance-in electromagnetic theory was made by Michael Faraday who, in 1831, discovered expetimeil.tally that a current was induced in a conducting loop when the magnetic flux linking the loop changed.+ The quantitative relationship

' There is evidence that Joseph Henry independently made similar discoveries about the same time.

270 TIME-VARYING FIELDS AND.MAXWELLgS EQUATIONS 1 7

between the induced emf and the rate of change of flux linkage, based on experimental observation, is known as Faraday's law. It is an experimental law and can be considered as a postulate. However, we do not take the experimental relation concerning a finite loop as the starting point for dcvcloping the theory of clcctromagnctic induction. Instead, we follow our approach in Chapter 3 for electrostatics and in Chaptcr 6 for magnetostatics by putting forth the following fundamental postulate and devel- oping from it the integral forms of Faraday's law.

Fondnmmtal Postulate for Elrctramngnctic Induction

Equation (7-2) is valid for any surface S *it11 a bounding contour C, whclhcr or not a physical circuit exists around C. Of course, in a field with no time variation, dB/ir = 0, Eqs. (7-1) and (7-2) reduce, respectively, to Eqs. (3-5) and (3-8) for electrostatics.

In the following subsections we discuss separately the cases of a stationary circuit in a time-varying magnetic field, a moving conductor in a static magnetic field, and a moving circuit in a time-varying magnetic field.

7-2.1 A Stationary Circuit in a Tirne-Varying Magnetic Field

. For a stationary circuit with a contour C and surface S, Eq. (7-2) can be written as

If we define

YT = $c E . dl' = emf induced in iircuit with contour C (V). (7-4) r .)

(7-2)

:r o; not i& = ostatics. y c~rcuit ield. and

!, . k ;

, , i 1 $

1 and 1

8 = ds = magnetic flux :rooding surface S ( W b ) ; (7-5)

then Eq. (7-3) becomg i : . %

." I

Equation (7-6) states +at r11e electromotiue.fwce induced in u , v o t ~ o ~ a r y c l o ~ d circuil is q i ~ d l o t h t1~q(i/i1w r ~ o (I/ ~I ICIYYI,W of 111v I M I ~ ~ I V ~ ~ C /I11 K / I I I / , ~ ~ , ( , / / I ( , circ [ l i t . TIUS C, a \ l ; ~ t c n i c ~ l t oI' I : w d t / y ' \ I w o / ' c l o c / r o ~ t i o q ~ ~ e r ~ c i t l d ~ r / ~ o ~ i . A tiolc-r.~lc of ch'inge ol m;ignetic llux i~ iduces a11 electric licld according to Eq. (7-3), even in the abbence of r '

physical closed circuit.sThe negallve sign in Eq. (7-6) is an assertion that the induced emf will cause a turrent to flow in the closed loop in such a direction as to oppose the change in the linking magnetic flux. This assertion is known as Lm?s lair. The emf induced in a stationary loop caused by a time-varying magnetic field is a tronsjormer emf.

Exrmplc 7-1 A circular loop of N turns of cuhduming wire lies in the xy-plane with its center at the origin ofa magnetic field specified by B = a,B, cos (nrl2bJ sin wt , where b is the radius ofthe loop and o is the angqlar frequency. Find the emf induced in the loop.

Solution: The problem specifies a stationary loop in a time-varying magnetic field; hence Eq. (7-6) can be used directly to find the induced emf, Y The magnetic flux iinking each turn of the circular loop is

711' = Jfib [a,B, cos 26 sin w t . (1,271r dr ) I

; = ?f (i -.-I)Bo sin or.: n

Since there are N turns,:thc total flux linkage is NQ, and we obtain 1. ,

B,o cos ot (V). 71

. . The induced emf is seen to be ninety degrees out of time phase with the magnetic flux.

272 TIME-VARYING FIELDS AND MAXWELL'S EQUATIONS 1 7

0 0 0 0

0 0 df? --+ u @

Fig. 7-1 A conducting bar moving 0 0 @ B @ in a magnetic field.

7-2.2 A Moving Conductor in a Static Magnetic field

When a conductor moves with 'a velocity u in a static (non-time-varying) magnetic field B as shown in Fig. 7-1, a force F, = qu x B will cause the freely movable electrons in the conductor to drift toward one end of the conductor and leave the other end positively charged. This separation of the positive and negative charges creates a Coulombian force of attraction. The charge-sepa3tion process continues ~ ~ n l i l ( I I C ~'Ieclric a11d n\;lgwlic I~)I.~,CS 1~1I;111cc C:ICII 01llcr ; I I I O ;I sl;\lc' oI'c*(ll~ilil>rill~~l is rcachcd. A t ccluilib~.iu~n. w l ~ ~ c l i is tcnclicd vcry ropiclly, the ncl rorcc on tllc kcc cllargcs in the moving conductor is zero.

To an observer moving with the conductor, there is no apparent motion and the magnetic force per unit charge F,/q = u x B can be interpreted as an induced electric field acting along the conductor and producing a voltage

V2, =J: (U x B ) . dt'. (7-7)

If the moving conductor is a part of a closed circuit C, then the emf generated around the circuit is

This is referred to as a flux-cutting emf, or a motional emf. Obviously only the part of the circuit that moves in a direction not parallel to (and hence, figuratively, "cutting") the magnetic flux will contribute to V' in Eq. (7-8).

Example 7-2 A metal bar slides over a pair of conducting rails in a uniform magnetic field B = a,Bo with a constant velocity u, as shown in Fig. 7-2. (a) Determine the open-circuit voltage Vo that appears across terminals 1 and 2. (b) Assuming that a resistance R is connected between the terminals, find the electric power dissipated in R. (c) Show that this electric power is equal to the mechanical power required to move the sliding bar with a velocity u. Neglect the electric resistance of the metal bar and of the conducting rails. Neglect also the mqchanical friction at the contact points.

on and the .n induced

ted around

ily. the part iguratively,

m m n t i c ermine, the

I ' a . dissipated required to c mctol bar tact points.

F,ig. 7-2 A metal bar sliding over conducting rails (Example 7-2).

a) The moving bar generates a flux-cutting emf. We use Eq. (7-8) to find the open- circuit voltage Vo: ,

V, = V, - v2 -; gC (U x Bj 're

= J2! (a,u x a , ~ , ) . (a, 'it)

= - ~rB,h (V). (7-9) b) Wlicn a rcsis~ancc R ~i conneclcd between terminals 1 and Z,3 current I = icB,li R

will flow from terminal 2 to terminal I , so that the electric power. P,, dlsslpiited in R is

c) The mechanical power, P,,,, required Lo move the sliding bar is

P , = F . u (w), (7-1 1 )

where F is the mechanical force required to counteract the magnetic force, F,, which the magnetic field exerts on the current-carrying metal bar. From Eq. (6-159) we have

F,.= 1 S2t' dP x B = -a,IBoh (N). (7-12)

The negative sign in Eq. (7-12) arises because current I flows in a direction opposite to that of dP. Hence,.

F = - F , = axIBoh = a,u~;h' /~ (N). (7-13) -1 Substitut~on-of Eq. (7'13) in Eq . (7-1 1) proves P,,, = P,., which i~pholds tlic

principle of conservation of cncrgy.

E h m p l e 7-3 The Faroilny disk gerwrntor consists of n circular nietnl disk rot;ltmp with :l constant ongulilr vclocily (u in a ull~l'urm and constant magnetic lield of flux density B = a,B, that is parallel to the axis of rotation. Brush contacts are provided


1 a

I'

2

Ld Fig. 7-3 Faraday disk generator (Example 7-3).

at the axis and on the rim of thepdisk, as depicted in Fig. 7-3. Determine the open- circuit voltage of the generator if the radius of the disk is b.

Solutiorz: Let us consider the circuit 122'341'1. Of the part 2'34 that moves with the disk, only the straight portion 34 "cuts" the magnetic flux. ~ & a v e , from Eq. (7-8).

which is the emf of the Faraday disk generator. To measure V, we must ,lac n voltmeter or :I vcry high rcsisl;~nm so tI1i11 n o ;~pprcci:ihle current Ilowr i l l llle circuit lo modify thc cxlcrn;~lly t~pplicd n1:tgnclic licltl.

7-2.3 A Moving Circuit in a Time-Varying Magnetic Field

When a charge q mdves with a velocity u in a region where both an electric field E and a magnetic field L1 cxist, the clcctrornugnetic brcc 17 on y, us mcasured by a laboratory observer, is given by Lorentz's force equation, Eq. (6-5), which is repeated below:

F = q ( E + u x B). (7-15)

To an observer moving with q, there is no apparent motion, and the force on q can be interpreted as caused by an electric field E', where

:e open-

e a volt- 2 circuit

!c'iicld E :-ed by a repeated

on q can

t ' 4 4 il

Hence, when a conducting circuit with contbur C and surface S moves with a velocity u in a field (E, B), weme Eq. (7-17) in Eq. [7$) to obtain

sv:

\ dB

$ ~ ' ? d t ' * - L d t . d s + (u x B).dt' (V). t

(7-18)

Equation (7-18) is tHe general form of Faraday's law for a moving circuit in a time- varying magnetic fidd. The line integral on tge left side is the emf induced in the moving frame of t'&fbence. Thc first term o n thd right sidc represents the transformer emf due to the time inriation of B; and the second term represents the motional emf due to the motioli of the circuit in B. The division of the induced emf between the . transformer and the motimal parts depends on the chosen frame of reference.

Let us consider P circuit with contour C that moves from C, at time t to C, at limo r + AI in a chilngi~ig m;ignolic lield U. Thc motion may include trunslaticn, rotation, and distortibn in an arbitrary manner. Figure 7-4 illustrates the situaticn. The time-rate of chatlge of magnetic flux through the contour is

B(r + At) . ds, - J . ~ ( t ) d s , S 1

(7-19)

B(r + At) in Eq. (7-19) can be expanded as a Taylor's series:

a q t ) B(i + At) = B(t) + - At + H.O.T.,

at where the high-order tkrms (H.O.T.) coniain the second and higher powers of (Ai). Substitution of Eq. (7-20) in Eq. (7-19) yields

Fig. 7-3 A moving circuit time-varying magnetic field.

276 TIME-VARYING FIELDS AND MAXWELL'S EQUATIONS I 7

where B has been written for B(t ) for simplicity. In going from C, to C,, the circuit coven a region that is bounded by S,, S,, and S,. Side surface S3 is the area swept out by the contour in time At. An element of the side surface is

d ~ , = dP x u At. (7-22) We now apply the divergence theorem for B at time r to the region sketched in Fig. 7-4:

where a negative sign is included in the term involving ds, because outward normals must be used in the divergence theorem. Using Eq. (7-22) in Eq. (7-23) and noting that V B = 0, we have

L, B. h, - k, B . ch, = -Ar$Ju x B)-dP.

Combining Eqs. (7-21) and (7-24), we obtain

t i c'R - C 0 * ' k = (' - . (1s - (f, (a x R) . l i t .

t l r -.s ;/ ., (. - - --- which can be identified as the negative of the right side of Eq. (7-18).

If we designate

1' = $c El dP = emf induced in circuit C measured in the moving frame

Eq. (7-18) can be written simply as

which is of the same form as Eq. (7-6). Of course, if a circuit is not in motion. f ' reduces to Y7 and Eqs. (7-2.7) and (7-6) are exactly the same. Hence. Faraday's law that the emf induce? in a closed circuit equals the negative time-rate of increase of the magnetic flux linking a circuit applies to a stationary circuit as well as a moving one. Either Eq. (7-18) or Eq. (7-27) can he used to evaluate the induced emf in the general case. IIil high-impedance .~oltmcter is insenal in a conducting circuit, i t will read the open-circuit voltage due to electromagnetic induction whether the circuit is stationary or moving. We have mentioned that the division of the induced emf in Eq. (7-18) into trimsformcr and motioo;ll emf's is not stiiqac, but their sum is ;~lw;lys eqi~al to that computed by using Eq. (7-27).

In Example 7-2 (Fig. 7-2). we determined tho open-circuit voltage V, by using Eq. (7-8). If we use Eq. (7-27), we have

in Fig. , .

4 ' ' I

orn~nls . ' ; i ' . - noting . ; .

(7-27)

ion. Y " ~ y ' s law rease of movmg I

fin the t, it f l 1

: c i r ~ - .&

! c1nf in 6 I

i alwi - !

)y using I ' I

I

7-2 / F: WDAY~S LAW 6~ ELECTROMAGNETIC INDUCTION 277 ,

1: . . I

6 1

and

which is the s ake & h q . (7-9). Similarly, for thd ~ a r d d a ~ disk generator in Example 7-3, the magnetic Aux

linking the circuit. 1?2'341'1 is that whish passes through the wedge-shaped area 2'342'.

and

Example 7-4 An h by i v rectangular conducting loop is sltuatcd iil ;I changing magnetic field B = a,B, sin o)r. The normal of the loop initially n~;ikes an angle 1 w ~ l h a,, as shown in f )g. 7-5. Iiind thc induced emf in thc loop: (a) when the loop is at rest, and (b) when the loop rotates with an angular velocity w about the x-axis.

. . b (a) Pcrspectivc vicw. (b) Vicw from +x direction.

I:ig. 7-5 A rcvlllll~ullr c o ~ l d a c h y loop rolnli~lg ill n cliu~lpiag ~ ~ l i i ~ ~ l c ~ i c licld (Example 7-4).

278 TIME-VARYING FIELDS AND MAXWELL'S EQUATIONS / 7

Solution

a) When the loop is at rest, we use Eq. (7-6).

a , = J ~ . d ~

= (a,,Bo sin wt) . (anhw)

= Bohw sin wt cos a. - Therefore

where S = hw is the area of the loop. The relative polarities of the terminals are ' as indicated. If the circuit is qompleted through an external load, *Ku will produce

a current that will opposc thc changc in (I). b) When the loop rotates about the x-axis, both terms in Eq. (7-18) contribute:

the first term contributes the transformer emf Y; in Eq. (7-28). and the second term contributes a motional emf r,': where

\I-

an - w x (a,Bo sin 031 . (a, dx) =l[( ; ) - a. - w x (a$, sin cot)] . (a, dx) +s:L( s )

Note that the sides 23 and 41 do not contribute to Y, and that the contributions of sides 12 and 34 are of equal magnitude and in the same direction. If a = 0 at t = 0, then cr = wt, and we can write

f /" . ', = B,Sw sin wt sin wt . (7-29)

The total emf induced or generated in the rotating loop is the sum of f'. in Eq. (7-28) and V:, in Eq. (7-29):

\ - - ,

which has an ;tngulnr frcqucncy 20). We can determine the total induced emf Y: by applying Eq. (7-27) directly.

At any time t , the magnetic Run linking the loop is

@(t) = B(t) . [a,(t)S] = B,S sin o t cds cr

= BoS sin wt cos o t = iBoS sin 2o t .

;ids are xoduce

~trlbute: . second 7

butio ions x = O a t

(7-29)

of %< in

( 7 C )

1 1

I Jir& ,.

1 i r ,

' 1 - , 3-3 I MAXWELL'S EQUh-tONS 279 t

. e L 8 !

Hence r: . ( ' < * d@

i : y:F --= dl -- ri (; BOs sin 2wt

% I , , I . = - B,So cos 20r ,

as before.

The fundamental dbstulate for electromagn~tic induction assures us that a time- varying magnetic figld gives rise to an electric field. This assurance has been amply verified b y numerous npcrimcnts. The V x E = 0 equation in Table 7-1 must ' therefore be replacekl by Eq. (7-1) in the time-varying case. Following are the revised set of two curl and two divergence equations from Table 7-1.

In addition, we know that the principle of conservation of charge must be satisded at all times. The mathematical csprcssion of charge conservation is the equation of c o ~ ~ l i r l u i t y , I S q . ( 5 XI), which IS ~'cpc;~tctl Iwlow.

The crucial question here is whether the set of four equations in (7-31a, b, c, and d) are now consistefit +,with the requirement specified by Eq. (7-32) in a time-varying situation. That the Answer is in the negative is immediately obvious by simply taking the divergence of Ed. (7-31 b),

which follows from the null identity, Eq. (2-137). We are reminded that the divergence of lhc cusl-of-;in$ MI-hc11;wd vector field 13 zero. Sincc Eq. (7-32) :lsserts V . J ~ O C S 11ot v:111i~h ill L I ' I ~ I I I C - V ; I ~ ~ ~ I \ ~ S ~ ~ L I : I [ ~ O I I , 1 i 1 , ( 7 : ~ ) i,, s ~n ' gcllcr;~I, not true.

How should Eqfi. (7-31a, b, c, and d) be modified so that they are consistent with Eq. (7-32)? First of all, a term ap/St must be added to the right side of Eq. (7-33):

\-

Equation (7,361 indicates that a time-varying electric field will give rise to a magnetic field. even in the absence of a current flow. The additional term aD/& is necessary in order to make Eq. (7-36) consistent with the principle of conservation of charge.

It is easy to verify that 2DBt ha's the dimension ofa current density (SI unit: Aim2). The term SD/dt is called displacement e~rrrent density, and its introductibn in the V x H equation was one of the major contributions of James Clerk M;i\\vcll (1831-1579). In order to be cotisistent with tlie cqu;ltion of continuity in a time- varying situation. both of the curl equations in Table 7-1 muh'be generalized. The set of four consistent equations to replace the inconsistent equations, Eqs. (7-31a. b. c, and d), are

They are known as Maxwell's equations. These four equations, together with the equation of continuity in Eq, (7-31) and Lorentr's force equation in Eq. (6-5), form the foundation .of electromagnetic theory. These equations can be used to explain and predict all macroscopic electromagnetic phenomena.

Although the four Maxwell's equations in Eqs. (7-37a. b, c, and d) are consistent, they are not all independent. As a matter of fact, the two divergence equations, Eqs. (7-37c and d). can be derived from the two curl equations, Eqs. (7-37a and b), by making use of the equation of continuity. Eq. (7-32) (see Problem P.7-7). The four fundamental field vectors E, D, B, H (cach having three components) represent twelve unknowns. Twelve scalar equations are required for the determination of these twelve unknowns. The required equations are supplied by the two vector curl equations and the two vector constitutive relations D = aE and H = B/p, each vector equation being equivalent to three scalar equations.

(7-36)

lgnetic xssary large. 1/m2). in the dxwell

L 11 ,--- d. The 7-3'

7-374

7-37b)

7-37c)

7 -37d)

~ t h the (6-5),

!sed to

sistent, . ; i t i ~ - P

and u), 7). 7;' ?rese. .. ~f these or curl u, each

. ?

j . , 3 . " "I..

. *., - . ' . I:

.. ,+. .

, .

I

!

. . ' . 3

. 2

~,

i . , f i , i ,

I ; . . . ,' . , * - . , . .

, '

I

. , i

. , I /

iI ' 8

i ' . - , , 8 - 1 . 7-3.1 Integral Form o ~ , ~ ~ x w e l l ' s Equations; :

. . i 7 ' i 1 ' The four Maxwell's bq~a'lions in (7-37a b, c,.and d) are differential equations that

,are valid at ever$ $oint;h space. In explain/ng electromagnetic phenomena in a physical environm t, we must deal with dhite objects of specified shapes and boundaries. It is bo ,venient to convert the differential forms into their integral-form equivalents. We ta f e the! sdrface integral of both sides of the curl equations in Eqs. (7-37a) and (7~37b) over an open surface'! with a contour C and apply Stokes's theorem to obtaiti I . .

, , " . i)

I I E . d P * - .-•

s at ds (7-38a) I

. - and

-

Taking the volume ihtegdl of both sides'of the divergence equations in Eqs. (7-37c) and (7-37d) over a ~o lumk v with a closed surface S and using divergence theorem, we have

and

The set of !OL& equat~ons in (7-38a, b. c and d) are the integral form of Ma~well's t liq. (7-38a) is the same as Eq. (7-2), which is an expression ctrarnrignetic induction. Equation (7-38b) is a generalization

law given in Eq. (6-701, the latter applying only to static magnetic urrent density J may consist of a convection current density of ;I irec-charge distribution, as well ;a ;; canduction current t h presence of,an electric field in a conducting medium. The

s the:c~rrent I flowing through the open surface S. a n b e recognized as Gauss's law, which we used extensively icli remains the same in the time-varying case. The volume

ihtegral ofAp eqvals the total charge Q that is enclosed in surface S. No particular h w is associated with EQ (7-38d); but, incdmparlng it with Eq. (7-38c), we conclude that there are no isdlatrd magnetic charges and that the total outward magnetic

. .. _. " - - i ; > ,',""' > 2 : : . " .

? ' 282 TIME-VARYING FIELDS AND MAXWELCS EQUATIONS / 7

rh

Table 7-2 Maxwell's Equations :.+ r..fl

Differential Fonn Integral Form Significance

dB V X E = - -

at dt Faraday's law.

dD dD fC H . de = I + 6 - ds Amfirs's circuital law

V - D = p Gauss's law.

V . B = o ' $ B . L = o No isolated magnetic charge.

6

flux through any closed surface is zero. Both the differential and the integral forms of Maxwell's equations are collected in Table 7-2 for easy reference.

Example 7-5 An AC voltage source of amplitude V, and-angular frequency w, U. = V, sin wt, is connected across a parallel-plate capacitor C,. as shown in Fig. 7-6. (a) Verify that the displaccmcnt current in t l~c capacitor is the same us thc conduction current in thc wires. (b) Detcrminc the magnetic licld intensity at il distancc r from the wire.

Solution

a ) The conduction current in the connecting wire is

i c = c l 3 = c v dt 1 0 a cos at (A).

For a parallel-plate capacitor with an area A, plate sepmtion d, and a dielectric medium of permittivity e, the capacitance is 0

- 4 (

Fig. 7-6 A parallel-plate capacitor connected to an AC voltage source (Example 7-5).

1 forms

-I

acy o, ig. 7 luctioil r from

electric

I , .I(c < . I > . ; a%; ' : ;i ,

, ' 7-4 / POTENTIAL FUNCTIONS 283 ,

;1 : , 1 , I

1 t !-I , '

With a vdltage 0, npidating betweenthe plates, the uniform electric field intensity E in the dielectfic is dqual to (neglecting fringing effects) E = vJd, whence '

The displacemeht cupent is then t

I - = C,Vow cOs wt = i, Q.E.D.

b) The magnetic fleld ntensity at a distanck r from the conductin, 0 wire can be found by applyi!lg the generalized Amp~re'scircuital law, Eq. (7--3Sb), to contocr C in Fig. 7-6 Two t/pical open surfaces Vith rim C may be chosen: (1) a planar diak surfacc S, i (2) .t curved surface S2 $assing through the diclectrlc medlum. Symmetry arouhd tl e wire ensures a constant Hd along the contour C The iine integral on the feft s de of Eq. (7-38b) is

For the surface S , , c nly the first term on the right side of Eq. (7-38b) is nonzero because no chiUges ire deposited along the wire and. consequently, D = 0.

j', J . dr = ic.= k1 vow cos wt.

Since the surface S, )asses through the dielectric medium. no conduction current 8,. If the second surface idtegral were not there, the right side of ohld 11e zero. This would rdsult in a contradiction. The inclusion

of the displgcerhent current term by Maxwell eliminates this contradiction. As we have showti in ]]art (a), iD = i,. Hence we obtain the same result whether surface S , or sdkfacc S, is chosen. Equating the two previous integrals, we find

!n Section 6-3 the concept of the vector magnetic potential A was introduced because of the solenoidal nature of B (V B = 0):

. .

284 TIME-VARYING FIELDS AND MAXWELL'S EQUATIC$S 1 7 bc . s' 1'

If Eq. (7-39) is substituted in the differential form of faradiY9s law, Eq. (7-I), we get

Since the sum of the two vector quantities in the parentheses of Eq. (7-40) is cur!-free, ' it can be expressed as the gradient of a scalar. To be consistent with the definition of the scalar electric potential Vin Eq. (3-38) for electrostatics, we write

( I


In the static case, aA/dt = 0, and Eq; (7-41) reduces to E = - V V . Hence E can be determined from V alonc; and B, from A by Eq. (7-39). For time-varying fields, E depends on both V and A. Inasmuch as B also dcpcnds on A, E and B are coupled.

The electric field in Eq. (7-41) can be viewed as composed of two parts: the first part, -VV, is due to charge distribution p; and the second part, -dA/dt, is due to time-varying current J. We are tempted to find V from p by Eq. (3-56)

and to find A by Eq. (6-22)

However, the preceding two equations were obtained under static conditions, and V and A as given were, in fact, solutions of Poisson's equations, Eqs. (4-6) and (6-20) respectively. These solutions may themselves be time-dependent because p and J may be functions of time, but they neglect the time-retardation effects associated with the finite velocity of propagation of time-varying electromagnetic fields. When p and J vary slowly with time (at a very low frequency) and the range of interest R is small compared with the wavelength, it is allowable to use Eqs. (7-42) and (7-43) in Eqs.

'.- (7-39) and (7-41) to find quasi-static fields. We will discuss this again in subset-

. L -( tion 7-7.2. Quasi-static fields are approximations. Their consideration leads from field

theory to circuit theory. However, whcn'thc sourcc frequency is high and thc range

be get '

(7-40)

&free, ition of

(7-fl

. 1- . 1 Iicld\, C~uplcd. hc first tluc to

(7-42)

4

(7-43)

Ins, m d j (6-30)

I and J

I C d (* "P ' is sm*ll

in l! subsec-

Im field le range

7-4 1 "TENTIAL FUNCTIONS 285 : b, * ,

I' : > ,

of interest is no longer small in comparison to the wavelength, quasi-static solutions ' will not suffice. Tirhe-rethrdation effects mtilt then be included, as in the case of

electromagnetic radiation from antenn&.!~hese points will be discussed more fully ' when we study soldions to wave equations. i

Let us substitute Eqs. (7-39) and (7-41) into Eq. (7-37b) and make use of the constitutive relations H = B/p and D = EE. We have

where a homogeneous r-iediutn has been'assumed. Recalling the vector identity for V x V x A in Eq. (6-16a), we can write Eq. (3-44) as

d2A V(V A) - V'A = pJ - V

0 r i12A

V2.4 -pi----= -uJ + V (7-45) 3 t Z

Now, the definition of .L vcctor rcquircs the specification of both its curl and its tlivcrycncc. A I ~ I I O L I ~ ~ L I I L w1.1 o1.A is d~sig11;~~cd U i n ECI. (7 -30), wu arc slill :I[ libcrly to chnose the divergence of A. We let

which makes the se'konc~ term on the right side of Eq. (7-45) vanish, so we obtain I

Equation (7-47) is the l~o~~/w~noyolcous. w v e eyrrutiorl for vector. pot~rlrid A. I t IS

called a wave equatim bxause its solutions represent waves traveling with a velocity equal to I/,&. This will become clear in Section 7-6 when the solution of wave equations is discusseti. The relation between A and V in Eq. (7-46) is called the Lorcnt: corditior~ (or Lora~tz galye) for potentiat. It reduces to the condition V . A = O in Eq. (6-l9)+x$atic fields. The Lorentz condition can be shown to be consistent with the equation of continuity (Problem P.7'8).

A correspondini wave equation for the scalar potential V can be obtained by substituting Eq. (7-41) in Eq. (7-37c). We have

which, for a constant e, leads to

d P V 2 V + - ( V ' A) = --. (7-48)

at E

Using Eq. (7-46), we get I

which is the nonhomogeneous wave equatiun for scalar potential V . The nonhomogeneous wave equations in (7-47) and (7-49) reduce to Poisson's eqcations in static cases. Since the potential functions given in Eqs. (7-42) and (7-43) are solutions of Poisson's equations, they cannot:be expected to be the solutions of nonhomo, aeneous wave equations in time-varying iituations without modification.

7-5 ELECTROMAGNETIC BOUNDARY CONDITIONS -1

In order to solve electromagnetic problems involving contiguous regions of different constitutive parameters, it is necessary to know the boundary conditions that the field vectors E, D, B, and H must satisfy at the interfaces. Boundary conditions are derived by applying the integral form of Maxwell's equations (7-38a, b, c, and d) to a small region at an interface of two media in manners similar to those used in obtaining the boundary conditions for static electric and magnetic fields. The integral equations are assumed to hold for regions containing discontinuous media. The reader should review the procedures followed in Sections 3-9 and 6-10. In general, the application of the integral form of a curl equation to a flat closed path at a boundary with top and bottom sides in the two touching media yields the boundary condition for the tangential components; and the application of the integral form of a divergence equation to a shallow pillbox at an interface with top and bottom faces in the two contiguous media gives the boundary condition for the normal components.

The boundary conditions for the tangential components of E and H are obtained from Eqs. (7-38a) and (7-38b) respectively:

* )uCC )..L:l' .-<

r - - We note that Eqs. (7-50a) and (7-50b) for the time-varying case are exactly the same ! as, respectively, Eq. (3-1 10) for static electric fields and Eq. (6-99) for static magnetic

fields in spite of the existence of the time-Varying terms in Eqs. (7-38a) and (7-38b).

(7 - 49)

, I homo- . ; '

1 static Ions of 3ncous

i!ferc I u t the .ill\ are I(! dl to b ~ h ( s ~ i ~ l -

ntcgr:il 13. The ~cneral, bound- - ry con- rm of a m faces )orients.

btained

he same riagnetic (7-38b).

fi I'i ' h

The reason is that: intlettiq the height oflt6e flQt closed path (obcda in Figs. 3-22 and 6-17) approach zerd, the:area bounded by the path approaches zero, causing the surface integrals of akpt i n d a ~ / a t to vanish'

Similarly, the b&ndary conditions fot the normal components of D and B are obtained from EqS. P-385) and (7-38d); ,

"

I .

. , These are the same a$, re: p6ctively, Eq. (3-l'l3qj for static electric fields and Eq. (6-95) ,

for static magnetic %Ads Gecause we start from the same divergence equations. We can make thk follawing general (Gtements about electromagnetic boundary

conditions: (1) Tltc iarig~~&i conlporlent of an E jeld is continuoris across nn inter- .race; ( 2 ) Thc ~ungcnliul cunzponcnl cf an H Jielzl is di.sc~ontit~rlo~i.s ucross t r n intc f 'I' Lice where u surface current e&s, the a~nount' of disconti~~uity b e i y detcrtnineti by E y . (7-506); ( 3 ) The noimal component of a D jield is discontinuous across an inrcrjiuce

. where u surfucc dlurqe c.xisi,s., the umount oj' tliscontinuity hcing dctotn~tled hj. Eq. ( 7 - 5 0 ~ ) ; and (4) The normal component of a B jieltl is continuous across nn inrefice. As we have noted previoikly, the two divcrgcllcc equ;!lions can bc dcrivcd from the Lwo curl cqualions find Lllc crl\li~~ioti oScotili!l~rily: IICIICC, tllc ho t~~ i~ I ;~ ry conditions i n Eqs. (7 5 0 4 ; I I ~ ( 7 50d), wl~icl~ ;tr( ~ ? l w i ~ ~ c d [IX)IN t l~c d ive~~gc~~cc cqu;~t io~~s , cant101 be iti~1cpc11~1~ti1 I'rotn ~ I I O S C in Lqs. (7-50a) :m1 (7 --job), which arc obtained from the curl equatibns. As a matter of fact, in the time-varying case the boundary condition for the tangential component o f l ~ in Eq. (7-50a) is equivalent to that for the normal compodent of B in Eq. (7-50d), and the boundary condition for the tangential component ot'H in Eq. (7-50b) is ekluivalent to that of D in Eq. (7-51)~). The simultaneous s$ecific'ation of the tangential component of E and the normal component of B at.. a boundary surface in a ,timevarying situation, for example, would be redundant,and, if we are not careful, could result in contradictions.

We now examine the.important special cases of (1) a boundary between two lossless l i n e ~ r medk, and (2) a boundary between a good dielectric and a good

'C . conductor. - .

7-5.1 Interface'twtween two ~Qss~ess . Linear Media

A lossiess linear medium can be specified by t i permittivity E and a permeability p, with a = 0. There are usually no free charges and no surface currents at the interface between two lossless media, v e set p, = 0 and J, = 0 in Eqs. (7-50% b, c, and d) and obtain the boundary conditions listed in 'Table 7-3.


Table 7-3 Boundary Conditions between Two Lossless Media

1

(7-51a) I

i

7-5.2 Interface between a Dielectric and a Perfect Conductor

A pcrl'cct C O I I I ~ I I C ~ O S is o t i c with a11 i ~ l l i l i i l c L'OII~IIC~~V~L~. 111 llle physical world we only have "good" conductors such as silver, copper, gold, an'szluminium. In order

i to simplify the analytical solution of field problems, good conductors are often con- 1

I sidered perfect conductors in regard to boundary conditions. In the interior of a perfect conductor, the electric field is zero (otherwise it would produce an infinite 1 current density), and any charges the conductor will have will reside on the surface only. The interrelationship between (E, D) and (B, H) through Maxwell's equations

! ensures that B and H are also zero in the interior of a conductor in a time-varying

i i

s i t ~ a t i o n . ~ Consider an interface between a lossless dielectric (medium I) and a perfect conductor (medium 2). In medium 2, E2 = 0, M2 = 0, D2 = 0, and B, = 0. The

I I f 1 I

Tablc 7-4 Boundary Conditions between a Dielectric (Medium I) ; l

and a Perfect Conductor (Medium 2) (Time-Varying Case) i

On the Side of Medium 1 On the Side of iMedlurn 2

El, = 0 E2, = 0 (7-52a)

(7-52~) -\

an2 ' DI = P s D,, = 0

B , , = 0 R,, = 0 (7-52d) t -

' In the static case, a steady current In a conductor produces a static magnetic field that does not affect the electric field. Hence, E and D within a good condudtor may be zero, but B and H may not be zero.

orW-3 .n oruer cn iOr 0 1 ;i

infinite \urface

jumons - L aryirtg .d a per- = 0. The

(7-52a)

(7-52b)

(7P.c)

(7-, - 4 ) I -

2s not affect L be zero.

1

C I general boundary ctmditibns in Eqs. (7-50, b, c, and d) reduce to those listed in Table 7-4. When \uj apply Eqs. (7-52b),and ('I-52c), it is important to note that the reference unit nokmdl is ail outward normal /ram m d i u m 2 in order to avoid an error in sign. As mentionM in Section 6-10, currents in media with finite conductivities are expressed in tenhs orvolume current denjities, and surface current densities defined for currents flohing through an infinitesimal thickness is zero. In this case, Eq. (7-52b) leads to thq condition that the tAngCntia1 component of H is continuous across an interface $th a conductor having a finite conductivity.

. ,

1

Exan~ple 7-6 The k arid H field of a certain propagating mode (TE,,) in a cross section of an a by b riectallgular waveguide are k = a,E, and H = a,H, + a,H., where

a IIX E, = - j o p - H, sin -

II a (7-534

where H,, w, p,, and /3 are constants. Assumiqg the inner walls of the waveguide are perfectly conductin& determine for the four inner walls of the waveguide (a) the surface charge densities m d (b) the surface current densities.

I

Solution: Figure 777 sliows a cross section of the waveguide. The four inner walls are specified by x = 0, x = a, y = 0, and y = b, The outward normals to these walls (medium 2) are, respkcti~ ely, a,, -a,, a,,and -a,.

Fig.. 7-7 Cross section of a rectangular x waveguide (Example 7-6).

290 TIME-VARYING FIELDS AND MAXWELYA'EQUATIONS 1 7

. a) Surface charge densities-Use Eq. (7-52c):

ps(x = 0) = a , . E,E = 0

a nx PAY = 0) = a), . r ,E = E , E,, = - j o p r , - H , sin - 71 a

a nx PAY = b) = - aye = - r o E y = j o p o - H, sin - n a

= -ps(y = 0).

b) Surface current densities - Use.Eq. (7-52b):

nx n 7t.Y = a,H, cos - - a, j p - H , sin - a * n a

nx a nx = -a ,H, cos - + a z j p - H , sin - a n (I

In this section we have discussed the relations that field vcctors must satisfy at an interface between different media. Boundary conditions are of basic importance in the solution of electromagnetic problems because general solutions of Maxwell's equations carry little meaning until they are adapted to physical problems each with a given region and associated boundary conditions. Maxwell's equations are partial differential equations. Their solutions will contain integration constants that are determined from the additional information supplic$ by boundary conditions so that each solution will be unique for each given problem.

7-6 WAVE EQUATIONS AND THEIR SOLUTIONS

At this point we are in possession of the essentials of the fundamental structure of electromagnetic theory. Maxwell's equations give a complete description of the

@tt l

" ' relation betwcen elcctromagnctic fields and chsrgs and curmnt distributions. Thcir . solutions provide the answers to all electromagnetic problems, albeit in some cases

the solutions are difficult to obtain. Special 'analytical and numerical techniques

mr thc

i I

7-6.1 Sol I

! for Potenti

I

L\'< for d c an(

Po of 1 '

Eq by of wi' soi

Eq - 11.

/-'

satisfy at portance laxwell's :ach with :e partial that are ltions so

A

uctur. f )n of the Ins. Their m e cases :cliniques

! 7- * -) WAVE EQUAT~ONS AND THEIR SOLUTIONS 291 I 1 .

t

may be devised to aid in the solution proceduie; but they do not add to or refine the fundamental structure. Such is the importanbe of Maxwell's equations.

, For given charge $nd current and J, we finf solve the nonhomogeneous wave equations, Eqs. for potentials A and V. With A and V determined, E dild Bacan be found ffbml.kespectively, Eqs. (7-41) and (7-39) by differentiation. 4

t .

L ' 7-6.1 Solution of Wave ~ ~ d t i o n s for Potentials I: I

. I I : . .. .

Wc now 'cokidci the'kol~hion of the nonbom&encous wave cqu:ltion. Eq. (7-49,, for ~cilliir electric potenti.~l V. Wc can do thisby first finding the solution for an ,

clcmental point oliarpc ili timc i , p ( / ) Ad, locticd at tbc origin ol' tile coordinates and lhen by summing the elkcts of all 111e charge elements in a given region. For a point charge at tfie origin, it is most convenient tb use spherical coordinates. Because of spherical symmetry, V depends only on R and t (not on 6' or 9). Except at the origin. V satisfies the following homogeneous equation:

1 d --(,2?\ - d2V ,

R2 iiR dR) ,LlE ---7- = 0.

dt-

We introduce a new variaole I

V(R, t ) = - U(R, t), R

which convcrts Eq. (7-54) to a2u z2u -- dR2

/ l E ---- = 0. 2z

Equation (7-56) is a one-d~mensional homogeneous wave equation. It can be verified by direct substitution (see Problem P.7-15) that any twice-differentiable function of (1 - ~fi) or of (t + R \,z) is a solution of~Eq. (7-56). Later in this section we will see that 3 f ~ n ~ t i d n o i l t + R&) does not correspond to a physically useful solution. Hence w: have ,

Equation (7-57) represents a wavestraveling m the positive R direction w t h a veloclty 1/42. As we see, the Function at R + AR at a later time i + At is . 1-

U(R + AR, i + Arj = f [r + At - ( R + AR)&] = jlr - ~ ~ 1 % ) .

Thus, the hnction retains its form if At = AR& = AR/u, where u = 1/& is the velocity of propagation, a characteristic of the medium. From Eq. (7-55), we get

To determine what the specific function f(t - R/u) must be, we note that for a I static point charge p( t ) Av' at the origin,

Comparison of Eqs. (7-58) and (7-59) enables us to identify !. p(t - Rlu) Au' i, A f (t - R/u) = I 6

4nc f

The potential due to a charge distribution over a volume V' is then f I

I Equation (7-60) indicates that the scalar potential at a distance R from the source at time 1 dcpe~ids on tlle vdue of the c l l q e dcllsity at an scs./icr tilne (i - Rle). It , takes lime R ~ I for the ellect of p to be fclt at dis t~~nce R:.Fw. this rcnson I,'(R, r ) in Ell. ( 7 - 0 ) i d c i l l / I I . I t is IIOLV clcal. Illat a function of ( t + R/u) cannot be a physically useful solution. since it would lead to the impossible 1

situation that the effect of p would be felt at a distant point before it occurs at the j source. I

The solution of the nonhomogeneous wave equation, Eq. (7-47), for vector 1 magnetic potential A can proceed in exactly the same way as that for V. The vector ; equation. Eq. (7-47), can be decomposed into three scalar equations, each similar to Eq. (7-49) for V. The retarded vector potential.is, thus, given by I

The electric and magnetic fields derived from A and V by differentiation will : obviously also be functionspf (t - Rlu) and, therefore, retarded in time. It takes time for electromagnetic waves to travel and for the effects of time-varying charges and / currents to be felt .at distant points. In the quasi-static approximation, we ignore i this time-retardation effect and assume instant response. This assumption is implicit I in dealing with circuit problcms.

i n

i I 7-6.2 Source-Free Wave Equations i 717 -

In problems of wave propagation we are concerned with the behavior of an electromagnetic wave in a source-free region where p and J are both zero. In other words. we are often interested not so much in how an electromagnetic wave is originated. l ~ t in how il propag;;ltcs. Ifllic wave is iu a sinlpi,c(li~lcar, iwtropic, il~ld homogeneous) t

(7-59) '

( 7 - 60)

: ',i)llrLC

l i , f -Y (it. r j in ctlc' f

ipo\slole 1 5 ,it the

) r v c C t O r

IK ~ec to r h slmllar

(7-61)

.ltion wdl

.aka time arges and we Ignore is i m R t

I

:in clcctro- Ilcr words. or!gmated, logeneous)

, r . ,

nonconducting rnedidL cihracterized by and p (0 = 01, Maxwell's equations (7-37a, b, c, and d) reduce to:

Equations (7-62a, b, c, and d) are first-order diffetcntlal equations in the two vnr~ables E and H. They can bd combined to give 3 second-order equation in E alone. To do this, we take the curl of Eq. (7-62a) and use Eq. (7-62b):

Now V x V x E = V(V - E) - V'E = -V2E because of Eq. (7-62c). Hence, iic hi^^

or, since u = I/&

In an entirely similar way we also obtain an equation in H: I

Equations (7-64) and (7-54) \ire homogeneous vector w a w equatrons. We C X ~ see t h a ~ ih Cartesian coordinatcs Eqs. (7-64) and (7-65) can each be

decomposed into three on;-dimensional, homogeneous, scalar wave equatlons. Each component of E arid of H wdl satisfy an equation exactly like Eq. (7-56), whose so1ut1p.s represent whves. We will extensively discuss wave behavior in varlous envirpnmerm in,the next two chapters.

7-7 TIME-HARMONIC FIELDS

Maxwell's equations and .III the eqo;~tionodcsived from thum so far i n this chapter hold hr clwlso~aignclic qu;mlilics with .lo ;trbi[r:~ry ~~nru -dape~~dc~ lc~ . . '1.11~ aclu:d type of time functions that the field quantities assume depends on the source functions p and J. In engineering, sinusoidal time functions occupy a unique position.


They are easy to generate; arbitrary periodic time functions can be expanded into Fourier series of harmonic sinusoidal components; and transient nonperiodic functions can be expressed as Fourier integralst Since Maxwell's equations are linear differential equatibns, sinusoidal time variations of source functions of a given frequency will produce sinusoidal variations of E and H with the same frequency in the steady state. For source functions with an arbitrary time dependence, electrodynamic fields can be determined in terms of those caused by the various Trequency cuniponcnts of the source functions. The application of the principle of superposition will give us the total fields. In this section wc examine time-hurrnor~ic (stcady-state sinusoidal) field relationships. *

The Use of Phasors - A Review

For time-harmonic fields it is condenient to use a phasor notation. At this timc we digress briefly to review the use of phasors. Conceptually it is simpler to discuss a scalar phasor. The instantaneous (time-dependent) expression of a sinusoidal scalar quantity, such as a current i, can be written as either a cosine or a sine function. If we choose a cosine function as the rrfel+rizce (which is usua1l:v'dictated by the func- tional form of the cxcitntion), thcn all derived results will rcfer to the cosine function. The specification of a sinusoidal quantity requires the knowledge of three parameters: amplitude, frequency, and phase. For example,

i(t) = I cos (wt + 4), (7-66)

where I is the amplitude; w is the angular frequency (rad/s)--c:, is always equal to 2nf, f being the frequency in hertz; and cb is the phase referred to the cosine function. We could write i( t) in Eq. (7-66) as a sine function if we wish: i(t) = I sin (wt + 47, with 4' = 4 + 7112. Thus it is important to decide at the outset whether our rcfcrence is a cosine or a sine function, then to stick to that decision throughout a problem.

To work directly with an instantaneous expression such as the cosine function is inconvenient when differentiations or integrations:of i( t ) are involved because they lead to both sine (first-order differentiation or integsktion) and cosine (second-order differentiation or integration) functions and bccausc it is tedious to combine sine and cosine functions. For instknce, the loop equati6n for a series RLC circuit with an applied voltage e(t) = E cos wt is:

If we write i(t) as in Eq. (7-66), Eq. (7-67) yields .

(cot + 4) + R cos (wt + 4) +l sin (wt + 4) = E cos wt . (7-68) w c 1 ' D. K. Cheng, Analysis of Linear Systems; Addison-Wesley Publishing Company, Chapter 5, 1959

ar:. of

Exa A z :

odic func- are lineur given fre- 11cy in the odynamic mponcnts will give

inusoidal)

, llllle we dlsciuss a

da1 scaldr l l L l l r p I f [!LC IL, IC-

run!. 7.

Kln.-..As:

17 -66)

q u 2 I to r i l ~ l ~ l l o l l .

( , I / + (I,'), rcfcrence problem. functlon

l u x thcy nd-order bine sine cuit with

'767)

(7-63)

1959.

i f " 4

1.

, . ndcd into : '

, - I ? . ., 7-7 I TIME-HARMONIC FIELDS r ~ i , t

7 4 ,?' ' i . 4

Complicated mathe~dticalimanipulation~ ,$re iequired in order to determine the unknown I and 4. It 1' . +I 1 ' C

, It is much sim$ef,ho u& exponential fGnctidns by writing the applied voltage as ' ii i '

e(t) = E cos wt = &[(EejO)yjw] \ = %(E,eJWL) , (7-69)

and i ( t) in Eq. (7-66) & . I C . 1 J ; i(t) = 9& [(IeJ@)eju']

I f: . :i = We (I,ejO'), (7-70)

where .*i rncnns ' U c ;cal &rt of." In Eqs.'(7-69) and (7-701, , I

are (scalar) phasors that contain amplitude and phase information but am independent of r . The phasor E, in Bq. (7-71a) with zero phase angle is the reference phasor. Now,

I

Substitution of Eqs. (7-69: through (7 7%) in Bq. (7-67) yields

, R f j oL-- I ,=E , , [ (' J l from which the current ph.rsor I, can bc solvcd clslly. Note that thc time-dcpendcnt factor eJ'"' disappears from Eq, (7'-T! because it is present in every term in Eq. (7-67) after the substitution dnd ,s i i i c r t k . ~ i h i ~ ~ l e d . ?his is the essence of the usefulness of phasors ir. the analysis O i linear ~ y s i c n ~ s ui!h time-harmonic excitations. After I, has beer, determined, the instantaneous current response i(t) can be found from Eq. (7770) by (1) multiplying I , by eJ", and (2) taking the real part of the product.

If' the applied voltage had been given as a sine ]u~~rt ion such as e ( t ) = E sln ojr.

the series RLC-circuit broblem would be solved in terms of phasors in exactly the same way; . q y the idstantaneous expressions would be obtained by taking the iiriqinory part ofthe product of the phasors with i>j("'. The complex phasors represent the magnitudes and the phase shifts of the quantities in the solution of time-harmonic problems.

Example 7-7 Expres9 3 cos wt - 4 sin wr us first (a) A , cos (wr + O,), and then (b) A, sin (or + 0,). Detenhina A, , d l , A, , and 0 , .

296 TIME-VARYING FIELDS AND MAXWELL'S EQUATIOW?%f 7

Solution: We can conveniently use phasors to solve this problem.

a) To express 3 cos o t - 4 sin o t as A , cos (o t + 01), we use cos o t as the reference and consider the sum of the two phasors 3 and -4e-j"I2(=j4), since sin wt = cos (o t - n/2) lags behind cos cot by n/2 rad.

Taking the real part of the product of this phasx and ej"', we have ,-

3 cos o t - 4 sin o t = 9s[(5ej53.1')ejw']

= 5 cos iwt + 53.1"). (7-74a)

So, A, = 5, and 0, = 53.1" = 0.927 (rad). b) To express 3 cos wt - 4 sin ot'as A , sin (ot + I ) , ) , we use sin wt as the reference

and consider the sum of the two phasors 3ejnI2 ! =j3) and - 4. j3 - 4 = ~~j1an-l 3/(-4) = 5ej143.1'

(The reader should note that the angle above is 143.1°;-no~36.9".) Now we take the imaginary part of the product of the phasor above and ejw' to obtain the desired answer:

3 cos o t - 4 sin wt = ~m[(5e~'~~. ' ' )e '" ' ] = 5 sin (ot + 143.1"). (7-74b)

Hence, A, = 5 and 0, = 143.1" = 2.50 (rad).

The reader should recognize that the results in Eqs. (7-74a) and (7-74b) are identical.

Time-Harmonic Electromagnetics

Field vectors that vary with space coordinates and are sinusoidal functions of time can similarly be represented by vector phasors that depend on space coordinates but not on time. As an example, we can write a time-harmonic E field referring to COS C0tt US

E(x, y, Z, t) = .%k [E(x, y, z) ejw'] , (7-75)

where E(x, y, z) is a vector phasor that contains information on direction, magnitude, and phase. Phasors are, in general, complex quantities. From Eqs. (7-75), (7-70), (7-72a), and (7-72b), we see that, if E(x, y, z, t) is to bc rcprcscntcd by thc vcctor phasor E(x, y, z), then dE(x, y, z, t)/& and 1 E(x, y, z, t) dt would be represented by, respectively, vector phasors joE(x, y, z ) and E(x, y, z)/jw. Higher-order differentiations and integrations with respect to t would be represented, respectively, by multiplications and divisions of the phasor E(x, y, z) by higher powers of jo.

' If the time reference is not explicitly specified, it is custom;arily taken as cos cot.

i

.efermce ;in ot =

(7-74a)

,eJerence

' i o .Tue 3 0btc11n

(7 -74b)

:.lib) dlIC

s of tlme ~rdinates crrrny to

(7-75)

sgmtude, L i-f-x le vector cnta' ' '

Ikren,.~- by multi-

4 :I

, . a ' \ ?!-7 1 TIME-HARMONIC FIELDS 297

1

WC now,t&td timu-hsnnonic Maxwell's ciuiitions (7-37a, b, c, ;,nd d) in terms of vector field3hasors (E, H) and source phasors (p, J) in a simple (linear, isotropic, and homogenkoud) medium as follows. I'

Thc sp;~ce-codrdin:~td i l rg iments llavc been omitted for simplicity. Tlir f ~ t [hilt the same notations are used b r thc phasors as arc used for their corresponding time- dependent qunntities shc lild create l i t tle conft~sion. bec;lose ivc will deal almost exclusively with t i m e h - ~ n o o i c liclds (md tllercfore with phasors) in the rest of this book. When there is a na.d to distinguish an instantaneous quantity from a phasor, the time dependcnp of th:: instimtaneous quantity will be indicated erpiicitly by the inclusion of a f in its arqrment. P h a ~ o r q w ~ t i t i e s are not functions of i. It is useful to note that any quantity xntaining j most nccess;~rily be a ph:rsor.

The time-harmonic \rave equations for scalar potential V and vector potential A-Eqs. (7-49) and (7-47)- become. respectively,

and V ' A t- k 2 A = - p,J ,

whcrc (7-78)

is called the waven~mber. Equations (7-77) and (7-78) are referred to as sm~honio- geneous Helmholtz's e q l d x s . The Lorentz condition for potentials, Eq. (7-46), is now

The phasor soludbns o i Eqs. (7-77) and (7-78) are obtained from Eqs. (7-60) and (7-61) respectiydy :

Je-'tR =.& J" - ' R . dv' (Wb/m). I

298 TIME-'.'2RYING FIELDS AND MAXWELL'S EQUATIONS / 7

These are the expressions for the retarded scalar and vector potentials due to time- harmonic sources. Now the Taylor-series expansion for the exponential factor e - j kR

is

where k, defined in Eq. (7-79), can be expressed in terms of the wavelength 1 = u/f in the medium. We have

Thus, if

or if the distance R is small compared to the wavelength I., e - jkR can be aiproximated by 1. Equations (7-81) and (7-82) then simplify to the static expressions in Eqs. (7-42) and (7-43), which are used in Eqs. (7-39) and (7-41) to find quasi-static fields.

The formal procedure for determining the electric anhiiagnetic fields due to time-harmonic charge and current distributions is as follows:

1. Find phasors V(R) and A(R) from Eqs. (7-81) and (7-82).

2. Find phasors E(R) = - V V - jwA and B(R) = V x A.

3. Find instantancow E(R. t ) = 9, [ E ( R ) cj'"'] and B(R. t) = 9?, [B(R)~~" ' ] for a cosine reference.

The degree of difficulty of a problem depends on how difficult it is to perform the integrations in Step 1.

j

Source-Free Fields in Simple Media

In a simple, nonconducting source-free medium iharacterized by p = 0, J = 0, a = 0, the time-harmonic Maxwell's equations (7-76a, b, c, and d) become

V x E = - jwpH (7-85a) V x 1 1 = ~ o ) E E (7-85b)

V . E = O (7-85~)

V . H = O . , (7-85d)

Equations (7-85% b, c, and d) can be combined to yield second-order partial differential equations in E and H. From Eqs. (7--64) and (7-65), we obtain

V2E + k2E = 0 (7-86) and

V211 + li21,1 = 0, (7--87)

ninlated \ in Eqs. t l C fif-$. !s LibL .o

"'1 for a

-form the

t . - ?- ' TIME-HARMONIC FIELDS 299 1

. . 3 , ; , ,- . , f: which arc hornopnmu.~ pctor Helmholizk eguutiunr. Solutions of homogeneous Helmholtz's equations with various bouqdae conditions is the main concern of Chapters 8 and 10.

Example 7-8 Show that if (E, H) are sol&iohs of source-free Maxwell's equations in a simple medium characterized by 6 and a, then so also are (E', B'), where

In Eqs. (7-88a) and (7-~8b), o is an arbitrary angle, and q = is called rhe intrinsic impedulzce of the medium.

Solutios: We prove the st'atement by taking the curl and the divergence of E' aod H' and using Eqs. (7-,85a, b, 6, and d):

V x E ' = ( V % E) cos 2 -t )1(V x H)sin u

= ( -jw,uH) cos u i, q(JoeE) sin u

I 1 V x h' = -- (V x E) sin u + LV x H) cos u

YI 1

= -- (-jopH) sin u + (jweE) cos a 1; YI '

= jtuc(qH sin a + E cob a) = joeEf; (7-89b) I

V d k' = ( V . E) cos u + q(V. H) sin a = 0 ; (7-89c)

' 1 V . k t = - - ( V * ~ : ~ i n u ( ~ . ~ ) c o s o = o . ' 1

(7-89dj

Equations (7-S9a, b. c, anJ ?J are source-free ~ v i ~ ~ r e l l ' s equations in El and HI,

This example shbws that ~~urce- f ree M^xwell9s equations for free space are invariant under the linear ,trogrf?rmation spedified by Eqs. (7-88a) and (7-8Sbl. An interesting special case is for ii = 2712. Equations (7-88a) and (7-88b) become

H ' = -r: 7 , (7 -90b)

Equations (7-90a) and (7-90b) show that $(E, 8) are soiutioss of source-jke Max- well's equations then so also are (E' = qH,.Hf = -Eh). This is a statement of the principle of duulity. his principle is a consequence of the symmetry of source-free Maxwell's equations.

300 TIME-VARYING FIELDS AND MAXWELL'S EQUATIONS / 7 .

83

If the simple medium is conducting (a # 0), a current J = a E will flow, and Eq. (7-85b) should be changed to

= jw€,E with

where E' = E and E" = a/w. The other three equations, Eqs. (7-85a, c, and d), are unchanged. Hence, all the previous equations for nonconducting media will apply to conducting media if E is rcplaccd by the cor~plcs p~wl ir t iv i ty E,. The rcal wuvc- number k in the Helmholtz's equations, Eqs. (7-56) and (7-87), will have to be changed to a complex wavenumber kc = o a .

The ratio E"/E' measures the magnitude of the conduction current relative to that of the displ ;~~cn~cnt CUI-I-~111. It is c ; I I I c ~ :I loss t l ~ i ~ ~ q c n t ~ C C ; I U S C i t is a measure of thc ohmic loss in the medium: --. --.

E" a tan 6 , = -= -.

E' ..WE

The quantity 6, in Eq. (7-93) may be called the loss angle. A medium is said to be a good conductor if a >> OK, and a good insulator if WE >> a. Thus, a material may be a good conductor at low frequencies, but may have the properties of a lossy dielectric at very high frequencies. For example, a moist ground has a dielectric constant 6 ,

and a conductivity o that are, respectively, in the neighborhood of 10 and lo-' (S/m). The loss tangent ajoe of the moist ground then cquals 1.8 x lo4 at 1 (kHz), making it a relatively good conductor. At 10 (GHz), a/oc becomes 1.8 x and the moist ground behaves more like an insulator.?

Example 7-9 A sinusoidal electric intensity of amplitude 50 (Vjm) and frequency 1 (GHz) exists in a lossy dielectric medium that has a relative permittivity of 2.5 and a loss tangent of 0.001. Find the average power dissipated in the medium per cubic meter.

Solution: First we must find the effective conductivity of the lossy medium:

CJ tan 6, = 0.001 = -

OEoEr

' Actually the loss mechanism of a dielectric material is a very complicated process, and the assumption oTa constant conductivity is only a rough approximation.

T

REVIEW C

R. Cl'

R. :c.

R.'

u.- I < .

K.' -n I

R.7

R.7

R.7

R.7 Pro

R.7

R.7

R.7

R.7 110.

R.7

R.7 t l J1

7 R.7

, - -7rr < ,

- R.7 OOlT

R.7 t l r n ~

H.7

Ind Eq.

(7-91)

(7-92)

d), are 1 apply i wave- : to be

l u Illat 3 0 1 f i

r

1 0 he a ~y ?x a

C I C S ~ ~ I C 3tdllt E,

' tS/m). naking

moist - quency 2.5 and r cubic

/?

1

-

unlption

REVIEW QUESTIONS 301 ,

i d , i + Z ., a t ; - t

The average power di&ipadd per unit volb$e id * I

p i +JE& $aEZ :I?

=!+x (1.389 x loe4) x 50' =0.174(W/m3). '

REVIEW QUESTIONS ', 5

R.7-1 What constitute$ an'ekctromagnetostatic field? In what ways are E and B related in 3, conducting medium unddjr static conditions?

R.7-2 Write the fundatnental postulate for electromagnetic induction, and explain how it leads to Faraday's law.

R.7-3 State Lenz's law.

R.7-4 Writc thc expression for trxnsforrncr ern(.

R.7-5 Write the cxprestlion f i r llux-cuttlng emf.

R.7-6 Write the expression :or the induced emf in a closed circuit that moves in a changing rnagnctic field.

K.7-7 What is a Faraday d ~ s k generator'?

R.7-8 Wrlte the differential folm of Maxwell's quatlons.

7 A r c all four Maxwell's equations ~ndcpcndcnt'l Explain.

R.7-I0 Write the integral form of Maxwell's equatiohs, and identify each equation with the proper cxpcrimcntal law.

1

R.7-11 Explain the significance of displacement current.

R.7-12 Why are potentid functions used in electromaghetics?

R.7-13 Express E and B jn terms of po,ential funct~ons V and A.

R.7-14 What do we mean by ,l:/:rsl-\iuir~ f i ~ . r t s ? Are they exact solut~ons of Mauwell's q u ~ - tions'? Explain.

R.7-15 What is the Lorentz ciind~tion for pot.ent~als? what is ~ t s physical s~gn~ficance?

R.7-16 Write the n~nhoino~encous w,byc oq.~ation forlcalar potcnti:ll I/ ,und for vector porcn- Linl A. d

R.7-17 State%e-boundary ccmditions for the tangent\al component of E and for the normal component of B. -. R.7-18 Write the boundary conditions for the tangen,tjal component of H and for the normal coniponent of D.

R.7-19 Can a static magnetic field exist in the Interior of a perfect conductor? Explain. Can a time-varying magnetic field? Explain.

K.7-20 What do we mead by a retarded potential?

302 TIME-VARYING FIELDS AND MAXWELL'S EQUATIONS 1 7 w.;%

4& 3 <

~ . 7 - 2 1 In what ways do the retardation time and the velocity of wave propagation depend on the constitutive parameters of the medium?

R.7-22 Write the source-free wave equation for E and H in free space.

R.7-23 What is a phasor? Is a phasor a function o f t ? A function of w?

R.7-24 What is the diAerence between a phasor and a vector?

R.7-25 Discuss the advantages of using p h u o n in electromagnetics.

R.7-26 Write in terms of phasors the time-harmonic Maxwell*, equations for a simple medium.

,' R.7-27 Define wavenumber.

. .

R.7-31 In a time-varying situation how do we define a good c o n d & i g - ~ lossi dielectric?

R.7-32 Are conduction and displacement currents in phase for time-harmonic fields? Exphin.

PROBLEMS

P.7-1 Express the transformer emf induced in a stationary loop in terms of time-varying vector potential A.

P.7-2 The circuit in Fig. 7-8 is situated in a magnetic field

B = a= 3 cos (5x i07t - :rrx) (pT) . Assuming R = 15 (Q), f nd the currcnt i.

Fig. 7-8 A circuit in a time-varying

magnetic field (Problem P.7-2).

P.7-3 A conducting equilateral triangular loop is placed near a very long straight wire, in Fig. 6-37, with d = b/2. A current i(t) = I sin o r flows in the straight wire.

a) Determine the voltage registered by a high-impedance rms voltmeter inserted in the loop. b) Determins the voltmeter reading when the triangular loop is rotated by 60' about a

perpendicular axis through its center.

ipls medium. ,

tiais in terms

nconducting,

i t wire, snown

ud in the loop. 60 about a

, . r I ; *I I:, i PROBLEMS 303

. i

P.7-4 A conducthg Oircudr loop of s rudiuafl.1 (h) is rituatcd in the neighborhood of a very long power line carrying a 60-(Hz) current. a s shown in Fig. 6-38, with d = 0.15 (m). An AC milliammeter inserted the luop reads 0.3 (mA). Ansume the total impedance of the loop including the milliammeter to bqb.01 (0).

, . a) Find the mngdhde gf the current m ihe po&er line. b) To what angle about the horizontal axrs should the c~rcular loop be rotated in order to

reduce the milhammeter teading to 0.2 (mA)? , I

P.7-5 A conducting 01iding:bar oscillates oCer twd parallel conducting ra~ ls in a s~nuso~dally varying magnetic field

1 / , B = a, 5 css O J ~ (rnT),

as shown in Fig. 7-9. The posltion of the sliding bar is glven by x = O.ii(l - cos or) (m), and the rails are termmated in a resistance R = 0.2 (R). Find i.

R - Fig. 7-9 A conducting bar slldlng over parallel ralls In a tlme.varylng magnetlc field ( ~ r b b l e m P.7-5).

P.7-6 Assuming that a rcslstsnce R n connected across the slip rings of the rectangular conductlng loop that rotates in a constant magnetic field B = ap,, shown in Fig. 7-5. prove that the power dissipated ih R is equal to the power required to rotate the loop at an angular frequency o.

i

P.7-7 Derive the two divergence equations, Eqs. (7-37c) and (7-376). from. the two curl equations, Eqs. (7-3711 and 17-37b), and the equation of continuity, Eq. (7-32).

.I

P.7-8 Prove that the Lor:ntz condition for potentials as expressed in Eq. (7-46) is consistent with the equation of cdfitinxty.

P.7-9 Substitute ~ ~ 3 . (7-30) and (7-41) in Maxwell's equations to obtain wave equations for scalar potentiA V and vector potential A for a linear, isotropic but inhomogeneous medium.

P.7-10 Write the set of four Maxwell's equations, Eqs. (7-37a, b, c, and d), as elght scalar equations -'\ -

a' in Cartesian cdordi,l~tes, b) in cylindrical ~bordinates, c) in spherical coordinates.

P.7-11 Supply the detailed steps for the derivGion of the electromagnetic boundary conditions, Eqs. (7-50a, b, c, and d).


P.7-12 Discuss the relations

a) between the boundary conditions for the tangential components of E and those for the normal components of B,

b) between the boundary conditions for the normal components of D and those for the tangential components of H.

P.7-13 Write the boundary conditions that exist at the interface of free space and a magnetic material of infinite (an approximation) permeability.

P.7-14 The electric field of an electromagnetic wave

is the sum of

E, = ex 0.03 sin 10%

and

El = ex 0.04 cos [L08x(t - j) - 3 -'.

Find Eo and 8.

P.7-15 Prove by direct substitution that any iwice differentiable function of ( t - R&E) or of (t + R f i ) is a solution of the homogeneous wave equation, Eq. (7-56).

P.7-16 Prove that the retarded potential in Eq. (7-60) satisfies the nonhomogeneous wave equation, Eq. (7-49).

P.7-17 Write the general wave equations for E and H in a nonconducting simple medium where a ,charge distribution p and a current distribution J exist. Convert the wave equations to Hclrnholtz's cqu;ltions for sinusoidnl time dcpcndence.

P.7-18 Given thi1t E = a, 0.1 sin (IOnx) cos (6nlO"f - pi?) (V/ln)

in air. find M and /L

P.7-19 Given that H = a, 2 cos ( 1 5 ~ ~ ) sin (6x109t - pz) (Aim)

in air, find E and P. P.7-20 It is known that the electric field intensity of a spherical wave in free space is

Eo . E = a, - sin 0 cos (wr- kR). R

Determine the magnetic field intensity H.

P.7-21 In Section 7-4 we indicated that E and B can be determined from the potentials V and A, which are related by the Lorentz condition, Eq. (7-80). in the time-harmonic case. The vector potential A was introduced through the relation,B = V x A because of the solenoidal nature of B. In a source-free region, V . E = 0, we can definc another type of vector potential A,, such

PROBLEMS 305

those for the P.7-22 For a source-ffke me'dium where p ='O, J = 0, p = po, but where there is a volume density of polarization P; a single vector potentlal n, d a y be defined such that

I H = jm$, x 'n,. (7-94)

a) Express electricjfield iptendty E in temp of n and P. b) Show that x, satisfies the h~nhomo~enkous elmholtz equatlon

.I .; I-/

P V2n, + k2z a e ' - - --' (7-95)

€ 0

id a magnetic ' I ;. i

: I :

!

, !

P.7-23 Calculations cdhccrning thc clcctrornjlgnctit! clTect of currents in a good conductor usu:~lly ncglcct llic tlisplllcc~i~x~ll cu~.rcnl cvw ;iL ~ i i i~ rnw:~vc rrcqucncics. (a) Assunling t, = I and u = 5.70 x lo7 (S/m) for copper, compare thc magnitude of the displaccment current density with that of the conduction current density at 100 (GHz). (b) Write the governing diRerentia1 equation for magnetic field intensity H in a source-free good conductor.

;t'cnlials V and ase. The vector '

rnoidal nature ential A,, such

i

8 i Plane Electromagnetic Wqyes

8-1 INTRODUCTION

In Chapter 7 we showed that in a sburce-free simple medium Maxwell's equations, Eqs. (7-62a, b. c, and d) can be combincd to yield homogcncous vcctor wavi: equations in E and in H. These two equations, Eqs. (7-64) and (7-65). huvc exactly the same form. In free space. the sourcc-free wave equation for E is

-.. 1.

where

is the velocity of wave propagation (the speed of light) in free space. The solutions of Eq. (8-1) represent waves. The study of the behavior of waves which have a one- dimensional spatial dependence (plane waves) is the main concern of this chapter.

We begin the chapter with a study of thc propaption of time-harmonic plane- wave ficlds in an unbounded homogcncous nicdiurrl. Medium paramclcrs sirch xs intrinsic irnpcdance, attenuation constant, and phase constant will be introduced. The mcaning c ~ T . \ k i r i dcptll, tlic clcpth of wave pcncttxtion into :I good concluctor, will he explained. Electromagnetic waves carry with them clectromagnetic power. The concept of Poynting vector, a power flux density, will be discussed.

We will examine the behavior of a plane wave incident normally on a plane boundary. The laws governing the reflection and refraction of plane waves incident obliquely on a plane boundary will then be discussed, and the conditions for no reflection and for total reflection will be examined.

A uniform plane wave is a particular solution of Maxwell's equations with E (and also H), assuming the same direction, same magnitude, and same phase in infinite planes perpendicular to the direction of propagation. Strictly speaking, a uniform plane wave does not exist in practice, because a source infinite in extent would be required to create it, and practical wave sources are always finTte in extent. But, if we are far enough away from a source, the .wavefi.ont (surface of constant phase)

8-2 PLANE WAVES IN L~SSLESS MEDIA

In this and future focus our attention on wave behavior in the sinusoidal , , ,. great advantagd. The source-free, wave equation, Eq. '. a h ~ m o g e ~ e o u s ~ ~ e c t o r Helmholtz's equatioli (see Eq. . ,

: 7-86): I C

-P P;

(8-3) .

where k, is the free-bpace w~ver~umher :quatlons,

be a one-

* * .

In Cartesian cobrdiriates, Eq. (8-3) is equivalent to three scalar Helmholtz's -; equations, one each ih thc components E,, E,, and E,. Writing it for the component Ex, we have

Consider a uniform plane wave characterized by a uniform E, (uniform magnitude and constant phase) over plane surfaces perpenducular to z; that is,

aiL',/as2 = o and S2E,/ay2 = 0.

Equation (8-5) simplifies to

which is an ordinary differential equation becake Ex, a phasor, depends only on r. The solution of Eq. (8-6) is readily seen to be

E,(r) E; (z) + El(:) , = Eg+e-~bo~ + ~;&koz,

' -'.- (8-7) where ki and E; aie arbitrary (and,'in general, complex) constants that must be determined by b o u n d a j conditions. Note that since Eq. (8-6) is a second-order equation, its general s b l u t ~ m in Eq. (8-7) contains two integration constants.

Now let us examine what the first phesor term on the right side of Eq. (8-7) represents in real time. Using.cos wt as the reference and assuming E l to be a real

. . - 308 P~ANE ELECTROMAGNETIC WAVES 1 8 , : , ,

I ,

* .

0 z

Fig. 8-1 Wave traveling in positive z direction E:(z, t) = E,f cos (wt - k,~), for several values oft.

constant (zero reference phase af z = 0), we have

= E; cos (wt - koz) (V/&P (8-8)

Equation (8-8) has been plotted in Fig. 8+ for several values oft. At t = 0, E:(r, 0) = ,

E; cos koz is a cosine curve with an amplitude E;. At successive times, the curve ellcctively travels in the positive i direction. We have, then, a truuelitly wove. I f we fix our attention on a particular point (a point of a particular phase) on the wave, we set cos (tot - k,:) = a constant or

wt - koz = A constant phase, from which we obtain

dz w - - --- - c. dr k, (8-9)

Equation (8-9) assures us that the velocity of propagation of an cquiphase front (the phase velocity) in free space is equal to the velocity of light, which is approximately 3 x lo8 (m/s) in free space.

The quantity k , bears i definite relation to the wavelength. From Eq. (8-4), k , = 2nflc or

which measures the number of wavelengths in a domplete cycle, hence its name. An inverse relation of Eq. (8-10) is I

( 2 . c 1~ curve c. If we e wave,

mt (the mately

I ' , . 8-2 1 P ~ A N L WAVES IN LOSSLESS MEDIA 309 . . I I ' I . A 1 t

I i

. ' < ! ! i f" . ~ ~ u a t i o n s (8-10) adb (8711) are valid withoet the subscript 0 if the medium is a lossless material sucd as a.perfect dielectric. :

It is obvious hitbout replottins that thk sehond phasor term on the right side of 'Eq. (8-7), E;#O*, rdbresents a cosinusoidal wive traveling in the - .- direction with the same velocity e. Id an anbounded regiori wC are concerned only with the outgolng wave; hcncc, if tllc S(IIII.CC is on IIic ldl. Ibc:ncg~livcly goi~lg WiiVI. does not exist, and Eg = 0. However, if there are dmontinuities injhe medium, reflected waves traveling in the opposite direction must also be coniidered, as we will see later in this chapter.

The associated rbagne'tic field H can 65 folind from Eq. (7-85a) - -

,! :I, 1 ' '

I 0 0 1

which leads to

I aE+(z) H,+ =-A - j y ~ ~ dz

, H+ = O .

Thus, H,? is the only hon cero component of EI: and since

. - We have introduced a nej i quantity, q,, in Eq. (8-1)):

which is called the intrinsic in~ped~?dnnee of theflee qoce . Because ,lo is 3. real number H;(z) is in phase with ~ : ( r ) ; and we can write tlie instantaneous expression for H as ' Q(z, .) ;,H; (z, i) = i,, 9a[HT(r)ej"]

Hence, for a uniform plhne wave, the ratio o;the magnitudes of E and H is the intrinsic impedance of the medium. We a l ~ o ~ n o t e that H is perpendicular to E and that both are normal to the direction of propagation: The fact that we specified E = axEx

310 PLANE ELECTROMAGNETIC WAVES 1 8

is not as restrictive as it appears, inasmuch as we are free to designate the direction of E as the +x direction, which is normal to the direction of propagation a,.

Example 8-1 A uniform plane wave with E = axE, propagates in a lossless simple . medium ( E , = 4, p, = 1, cr = 0) in the + z direction. Assume that Ex is sinusoidal with

a frequency 100 (MHz) and has a maximum value of + low4 (V,/m) at t = 0 and 1 = (m).

a) Write the instantaneous expression for E for any t and z. b) Write the instantaneous expression for H. c) Determine the locations where E, is a positive maximum when t = (s).

Solution: First we find k.

a) Using cos ot as the reference, we find the instantaneous expression for E to be

E(z, t) = a,E, = a,10-4 cos (277 1O8t - kz + $).

Since E , equals + when the argument of the cosine function equals zero- that is, when

2rr 108t - k: + $ = 0,

we have, at t = 0 and z = $,

Thus,

E(Z, t ) = ax10-4 c ~ s 2~ io8t - -z + - -. ( 3 6

!

*= axlo-' cos [Zn lost - (i - :)I ( ~ / r n ) .

This expression shows a shift of a mere $ in the +z direction and could have been written down directly from the statement of the problem.

b) The instantaneous expression for H is

where

= 0 and !

(s).

/7 E e

s m o -

~uld have

n t

I , 8-2 1 PLAhE WAVES IN LOSSLESS MEDIA 311

I - 1 *

Hence, 3 . r 1 ,

H ~ Z , t ) = a,, cos [b IO'L - 7 (z - i)] (*/mi.

' c) At t = lo-', witequate the argument of tke cosine function to +2nn in order to make E, a maximumi

Examining this result more closely, we note that the wavelength in the given medium is

Hence, the positive naximum value of Ex occurs at

13 Z,,, =, - L hL (m). 8

The E and H fleldr are shown in Fig. 8-2 as functions of z for the reference time t = 0.

Fig. 8-2 E and H fields of a uniform plane wave at f = 0 (Example 8-1).


8-2.1 Transverse Electromagnetic Waves

We have seen that a uniform plane wave.characterized by E = a,E, propagating in the + z direction has associated with it a magnetic .field H = a y H y . Thus E and H are perpendicular to each other, and both are transverse to the direction of propagation. It is a particular case of a transverse electromagnetic (TEM) wave. The ph;l:or field quantities are functions of only the distance z along a single coordinate <.-,is. We now consider the propagation of a uniform plane wave along an arbitl~iry

, direction that does not necessarily coincide with a coordinate axis. The phasor electric field intensity for a uniform plane wave propagating in the

+ z direction is E(z) = Eoe-jkz, (5-16)

where E, is a constant vector. A morc gcncrnl form oT Eq. (8-1 6) is ,

E(-Y, yr =) = Eoe-lk+-lkd"~ki' (8-1'7)

It can be easily proved by direct substitution that this expression satisfies the homogeneous Helmholtz's equation, provided that --.

1-

/if + k; + k: = W ~ , D E . (8-15)

If we define a wavenumber vector as

and a radius vector from the origin

then Eq. (8-17) can be written compactly as

E(R) = Eie-~k ' R - - E e-jk%. R 0 (v/m), (8-21)

where a, is a unit vector in the direction of propagation. From Eq. (8-19) it is clear that

k, = k a, = ka, . a, (8-22a) ky = k a, = ka, a , (8 -22b)

k, = k . a, = ka; a, , (8-22c)

and that a,. a,, a , a , and a, . aZ are direction cosines of a,. Thc gcomctrical relations or a,, and I< arc illustrated in Fig. 8-3, from which

we see that

a, R = Length OP (a constant)

is the equation of a plane normal to a,, the direction of propagation. Just as z = Constant denotes a plane of constant phase and,uniform amplitude for thc wave in

, r . a * I ' " it , . . ?,, ! I ,

agating in ' "' ! 3

; E a n d H # . ! ! f propaga- I

1

'he phacor I 1

inate :i?;k.

, 1 .

ting i3 the . , L; 1

i (8-16)

the homo-

P I 8)

(8-19)

(5 -20)

(8-21)

3) it is clear

' (8 -22a)

(8-22b) (8 -224

fro??ich (\

Just rfs * = ihc wave in

Fig. 8-3 Radius vector and wave normal Y

Plane of constant to a phdk front phase (phase front) of a uniform plane wave.

Eq. (8-16), a,, . R = Constant is a plane of constant phase and uniform amplitude for the wavc in Eq. (8-21). In a charge-free region, Y - & = 0. As a result,

Hence Eq. (8-23a) can be written as

a,, - E , = 0. (8-23b)

Thus the plane-wave solution in Eq. (8-17) implies that E, is transverse to thc Jirec- tion of propagatioli.

The magnetic held associated with E(R) in Eq. (8-21) may be obtained from Eq. (7-85a) as

--------,

where

This is a consequence Of the iact that 7 E, = 0, where E, is a constant vector (see problem P.2-18).

314 PLANE ELECTROMAGNETIC WAVES / 8

is the intrinsic impedance of the medium. Substitution of Eq. (8-21) in Eq. (8-24) yields

It is now clear that a uniform plane wavc propagting in an :irbitmry direction. a,,, is a TEM wave with E I H and that both E and H arc t~o rn~a l to n,.

Polarization of Plane Waves

The polarization of a uniform planewave describes the time-varying behavior of the electric field intensity vector at a given point in space. Since the E vector of the plane wave in Example 8-1 is fixed in the x direction (E = a,Ex, where Ex may be positive or negative), the wave is said to be linearly polurized in the x direction. A separate description of magnetic-field behavior is not necessary. inasmuch as the direction of H is definitely related to that of E.

In some cases the direction of E of a plane wave at a given point may change with time. Consider the superposition of two linearly waves: one polarized in the x direction; the other polarized in the y direction and lagging 90" (or n/2 rad) in time phase. In phasor notation we have

where El, and E,, are real numbers denoting the amplitudes of the two linearly polarized waves.

The instantaneous expression for E is

= axElo cos ( o t - kz) + a,Ezo cos

In examining the direction change of E at a givcn point as t changes. it is convenient to set z = 0. We have

E(0, t ) = axEl(O, t ) + a,Ez(O, t ) = axElo cos ot + h,Ezo sin o t . (8 -28)

As wt increases from 0 through 4 2 , n, and 3x12 - the cycle at 2n - the tip of the vector E(0, t ) will traverse an elliptical direction. Analytically, we have

El(O9 t) COS Wt = ---

El,

4. (8-24)

,$--26)

:ction, a,,

3

!or of the the plane e posit~ve wparate

:ec:~on of ,P

L c; <C

0 1 .'l r n/2 rad)

(8-27)

o linearly

onvenierlt

CH8,

271 - ?e vise u.. :-

i 8-2 / PLAN& WAVES IN LOSSLESS MEDIA 315

I 1' 1

and n

.i ' h'!.tt,>ii " . . ' . I ' . L .I

E2(0,t) $ 1 7 sih o t = - E2 0

which leads to the foilowing equation for an ellipse:

(8 -29)

Hence E, which is tqe sum of two linearly polarized waves in both space and time quadrature, is ellipticully pobrired i f E20 # Ela. and in circelurly polilr~zed if Ezb = . El,. A typical polarization clrcle is shown in Fig. 8-4(a),

When E2, = ElQ1 thc instantaneous angle ct which E makes with the r-axis a t -. = 0 is I

which indicates that E lotates at a uniform rate with an angular velocity io in a countcrclockwisc directicn. When the fingers of the right hand follow the direction

Fig. 8-4 Polarization diagrams for sum of two linearly polarized waves in space quadrature at z = 0: (a) circular polarization, E(0, t) = E,,(a, cos or f a, sin at); (b) linear polarizhtion, E(0, t ) = (a,E,, + ayE2,) cos wr


' of the rotation of E, the thumb points to the direction of propagation of the wave. This is a right-hand or positive circularly polarized wave.

If we start with an E2(z), which leads El(z) by 90" (71.12 rad) in time phase, Eqs. (8-27) and (8-28) will be, respectively,

E(z) = axEl ,e-jkz + a Y J 'E 20 e-jk? (8-31) and

E(0, t) = a,Elo cos ot - a,,E2, sin ot. (8-32)

Comparing Eq. (8-32) with Eq. (8-28), we see that E will stiJJ. be elliptically polarized. If Ez0 = El,, E will be circularly polarized and its angle measured from the x-axis at z = 0 will now be -ot, indicating that E will rotate with an angular velocity w in a clockwise direction; this is a left-hand or negative circularly polarized wave.

If E,(z) and El(z) are in space quadrature but in time phase, their sum E will be linearly polarized along a line that makes an angle tan-' (E20/E,o) with ,the x-axis. as depicted in Fig. 8-4(b). The instantaneous expression for E at : = 0 is

E(0, t) = (n,Elo + a,E,o) cos tot. (8-33) -----. The tip of the E(0, t) will be at the point P , when wt = 0. Its magnitude will decrease toward zero as wt increases toward n/2. After that, E(0, t) starts to increase again, in the opposite direction, toward the poiit P , where wt'= n.

In the general case, E2(z) and El(:), which are in space quadrature. can have unequal amplitudes (E2, # El,) and can differ in phase by an arbitrary amount (not zero or an integral multiple of 7~12). Their sum E will be elliptically polarized and the principal axes of the polarization ellipse will not coincide with the axes of the coordinates (see Problem P.8-4).

Example 8-2 Prove that a linearly polarized plane wave can be resolved into a right-hand circularly polarized wave and a left-hand circularly polarized wave of equal amplitude.

Solution: Consider a linearly polarized plane wave propagating in t h ~ + z direction. We can assume, with no loss of generality, that E is polarized in the x direction. In phasor notation we have

E(z) = axl&c-jk', But this can be written as

E(z) = E,,(z) + E d 4 where

Eo E,&) = - (a, - ja,)e- jkz , , . 2 (8 -34a)

and " . i '

Eo E k ( ~ ) = 7j- (a,., + jay)e-~kz. (8 -34b)

8 -3 COh

of the wave.

phase, Eqs.

(8-31)

(8 -32)

y polarized. n the s-axis r velocity 01 ' M'NL'L'.

m E will be 1 rhs x-axis,

vtd into a xi wave of

: direction. iection. In

, ' I

8-3 / PLANE WAVES IN CONDUCTING MEDIA 317 3 C '. .r,

i C r

From previous dig$ussions we recognize thdt E&) in Eq. (8-34a) and E,,(r) in *; Eq. (8-34b) represent, iespectively, right-h&d and left-hand circularly

. waves, each havingian amplitude E,/2. ~he;btatement of this problem is therefore . , proved. The conte:se statement that the sum of two oppositely rotating circuiariy

polilrimd waves of equal amplitude is a linearly polarized wave i s of course, also true.

~. 8-3 PLANE WAVES lhj , ; CONDUCTING MEDIA ' T .

' ' I

In a source-free codducting medium, the homogeneous vector Helmholtz's equation to be solve is

where the wavenumber kc = w& is a complex number because s, = 6 - jd' is complcx. 2s dcfincd in Eq, (7792). The dcrlvations and discussions pcruin~ng to plane waves in a lossless medium in Section 8-2 can be modified to apply towave propagation in a cpndo:ting medium by simply replacing k with kt. However, in an effort to conform with t l e conventional notation used in transmission-line rheory. it is cusmn~:~ry to dcflii~. :I p s i r p : ~ g ; \ ~ i o ~ ~ const;ld/, 11, such th:\r

I -- I

S i n c ~ 7 is c~mplex, wc i:ritc, with the help of.& (7-92)

where u and /I are, resjactively, the real and imaginary parts of 7 . Their physical significance will be explained presently. For a lossless medium, 0 = 0, 1 = 0, and / I = k = t o &

The Helmhoitz's cqwtion, Eq. (8-X), bcco~nes

The solution of E ~ : (8-38), whi& corresponds to a uniform plane wave propagating in the + s direction, is .- 1- E = a,E, = a,E,e-Yz, (5-39)

where we have assbmed that the wave ii linearly polarized in the .x direction. The propagation factor e-"' can be written as a product of two factors:

E, = ~ , i ~ - " ~ ~ - j L ' ' ,

As we shall see, both x and j3 are positive quantities. The first factor, e-", decreases


1 ...

as z increases and, thus, is an attenuation factor, and a is called an attenuation con- I stant. The SI unit of the attenuation constant is neper per meter ( N P / ~ ) . ~ The second

factor, e-jt', is a phase factor; P is called a phase constant and is expressed in radians per meter (rad/m). The phase constant expresses the amount of phase shift that occurs as the wave travels one meter.

General expressions of a and in terms of o and the constitutive parameters-E, p, and a-of the medium are rather involved (see Problem P.8-6). In the following paragraphs we examine the approximate expressions for a low-loss dielectric and a good conductor.

8-3.1 Low-Loss Dielectric

A low-loss dielectric is a good but'imperfect insulator with a nonzero conductivity, such that E" << E' or a / w ~ << 1. Under this condition y in Eq. (8-37) can besapproxi- mated by using the binomial expansion.

from which we obtain the attenuation conitant -

and the: phasc constant

It is seen from Eq. (8-40) that thc attcnualion constant of a low-loss dielectric is a positive constant and is approximately directly proportional to the conductivity 0.

The phase constant in Eq. (8-41) deviates only very slightly from the value w f i for a perfect (lossless) dielectric.

The intrinsic impedance of a low-loss dielectric is a complex quantity.

Since the intrinsic impedance is the ratio of E x and H, for a uniform plane wave, the electric and magnetic field intensities in a lossy dielectric are, thus, not in time phasc, as they would be in a lossless medium.

. - - -- - -. . . -. - - Ncper is a dimensionless quantity. If s! = 1 (Np/mJ, thcn a unit wave amplitude dccreascs lo :I magnitude

t . - I (=0.368) as it travels a distance of I (m). In terms of'field intensities 1 (Np/m) equals 20 log,,t = 8.69 (dB/m).

ductivity, : ~pproxi-

lectdic is a uctivity a. due a&€ -

8-3 1 PLANE ~ A V E S IN CONDUCTING MEDIA 319 ! ;- 1 L

The phase vhocky up is obtained from the ratio o/b in a manner similar to that - 1 in Eq. (8-9). (8t41), we have "

, I - I (mls).

8-3.2 Good Conductor .. 1 ,

A good conducto~ih a medium for which C . c' or o / w >> 1. Under this condition we can neglect 1 in tomparison with the te rd o/ jw in Eq. (8-37) and write

i I + j '7 i j u & J g = ' ~ ~ = -=

Jz or ~ = + j j l j z ( l +j)&,

where we have used thc relations

and o = 2nf. Ecjudtion (8-44) indicates that il and /j for a sood conductor arc approximately equal an.1 both increase us $ and &. For a good conductor,

L I

The intrinsic impedi-nce of a good conductor is

which has a phase angh: of 15'. Hence the msgnctic field intensity lags behind the electric field intensity by 45".

The phase velocity in a good conductor is

which is proportiunul to ,/r and I/&. Consider copper as an example:

o =i .80 x lo7 (S/m),

%.-%. p = 47c x 10-'(H/m),

/ I , , 770 (~nis ) :It 3 ( M I 17).

\vhicl~ is :lboul laicc 1I1c vcl~cily of suuiid in ~ l i r il~ld is many ordors of magnitude ;lower than the velocity of light in air. The Wavelength of a plane wave in a good conductor is ,


For copier i t i 3 (MHz), i = 0.24 ( m d . As a comparison, a 3-(MHz) electromagnetic wave in air has a wavelength of 100 (m).

At very high frequencies the attenuation constant a for a good conductor, as given by Eq. (8:45), tends to be very large. For copper at 3 (MHz),

cc = Jn(3 x 106)(4x x 10-7)(5.80 x lo7) = 2.62 x lo4 (Nplm).

Since the attenuation factor is e-"', the amplitude of a wave will be attenuated by a factor of e- ' = 0.368 when it travels a distance 6 = l/a. For copper at 3 (MHz), this distance is (112.62) x (m), or 0.038 (mm). At 10 (GHz) it is only 0.66 (pm)-a very small distance indeed. Thus, a high-frequency electromagnetic wave is attenuated very rapidly as it propagates in a good conductor. The distance d through which the amplitude of a traveling plane wave decreases by a factor of e- ' or 0.368 is called the s k i ~ l depth or the depth ?f'yrnetrcdion of a conductor:

Since cc = p for a good conductor, 6 can also be written as

At microwave frequencies, the skin depth or depth of penetration of a good conductor is so small that fields and currents can be considered as. for all practical purposes, confined in a very thin layer (that is, in the skin) of the conductor surface.

Example 8-3 Thc clcotric Gcld inlcnsity ol'a lincariy polarized uniform plane wave propagating in the + z direction in sea water is E = aJ00 cos (107nt) (V/m) at z = 0. The constitutive parameters of sea water are 6, = 80, pr = 1, and a = 4 (S/m). (a) Determine the attenuation constant, phase constant, intrinsic impedance, phase velocity. wavelength, and skin depth. (b) Find ihc distance a: which ihc amplitude of E is 1% of its value-at z = 0. (c) Write the expressions for E(z, t ) and H(z, t ) at z =

0.8 (m) as functions oft .

Solution

o f = - = 5 x lo6 (Hz), 211.

I' 8-3 1 PL&>E WAVES IN CONDUCTING MEDIA 321

f ? . h I

uated by a MHz), this -

I I

6 (pm) -a attenuated I !

which the I calicd the

;onductor 1 purposes,

)lane wave m) at r = = 4 (S/m'). nce. phase ~plitude of r, 1) at z =

f Y 4 I f * ' ; 5 Hence we cart. b&! theLformulas for good c'ondiktors:

a) Attenuation consfant,

a = - = ,/5nlo6(4n10-!)4 = 8.89 (Np/m).

Phase constant, P = = 8.89 (rad/m).

Intrinsic impedatice, . !

+ j - %

= ( I + j) $T

x 106)(4n x lo-') 4

- - ne'"'" (R).

Phase velocity,

Skin depth, 1 1 S = - = - = Q 1 1 2 ( cc 8.89

. m). -.. ,

b) Distance z, at which the amplitude of wavd decreases to 1 9 o f I ~ S ' value at 2 = 0:

1 4.605 z , =-In 100 = --

ci 8.89 - 0.518 (m).

c) In phasor notation, E(=) = 3x10()e-eze-~/'=.

The instantaneous expression for E is

E(z, t ) = Wt [E(z)eTa']

1% * Bd [axlOOe-"zeJ('-~q] = axlOOe-az cos (or - 8;). ,

At r7 = 0.8 (m), we have

E(O.8, r ) = aJOOe-0.8a cos (107nt - 0.81~')

= a,0.082 cos (107nt - 7.11) (V/m).

We know that a uniform plant wave is a TEM Lave with E I H and that both are normal to the direct~on of wave propagadon a,. Thus H = a,,H,. To find

I

322 PLANE ELECTROMAGNETIC WAVES I 8

H(z, t ) , the instantaneous expression of H as a function of t, we must not make the mistake of writing H y ( z , t ) = E,(z, t ) / v , , because this would be mixing real time functions E,(z, t ) and H Z ( z , t) with a complex quantity q,. Phasor quantities E,(z) and H y ( z ) must be used. That is,

from which we obtain the relation between instantaneous quantities

For the present problem we have, in phasors. ,

Note that both angles must be in radians before combiniug. The instantaneous -- expression for H at z = 0.8 (m) is then

We can see that a 5-(MHz) plane wave attenuates very rapidly in sea water and becomes negligibly weak a very short distance from the source. This phenomenon is accentuated at higher frequencies. Even at very low frequencies, long-distance radio communication with a submerged submarine is extremely .difficult.

8-3.3 Group Velocity

In Section 8-2 we defined the phase velocity, up, of a single-frequency planc wavc as the velocity of propagation of a n equiphase front. The relation between u p and the phase constant, /3, is

I

For plane waves in a lossless medium, /3 = w@ is a linear function of w. As a conscqucncc, the phase velocity u , = 114; is a constant that is indcpcndcnt of frequency. However, in some cases (such as wave propagation in a lossy dielectric, as discussed previously, or along a transmission line, or in a waveguide to be discussed in later chapters) the phase constant is not a linear function of o; waves of different frcquencics will propagate with dircrcnt phxx vclocitics. Inasmuch as all information-bearing signals consist of a band'of frequencies, waves of the component

not make . lixing real quantities

n water :. This phe- Srequencies, IS extremely

plane wave .\.em r r , and

P ~q 4s :I

icpcnc- >t of sy dielectric, Je to be dis- w : waves of d much as all e component

1 . frequencies travel with different phase velocities, causing a distortion in the signal wave shape. The signitl "disperses." The phenomenon of signal distortion caused by r dependence of the pHase ielocity on frequency 1s called dispersion. Given Eq. (8-43) . we conclude that a lossy dielectric is obviously & dispersive medium.

An information-bearing signal normally has a small spread of frequencies (sidebands) around a high carrier frequency. Such a signal comprises a "group" of frequencies and forms,a wave packet. A group u~locity is the velocity of propagation of the wave-packet erlielope.

C~ns ide r the simblest case of a wave packet that consists of two traveling waves having equal amplidde and s!ightly different angular frequencies wo + Aw and wo - A o (Aw << w,). The phase constants. being functions of frequency. will also be slightly different. Let the phase constants corresponding to the two frequencies be Do + A P and Po - AD. We have

E(:, t ) = = E~ cos [(coo - Aw)t - (Po + Ail):]

+ E g cos [((oO - Ao))t - (/j0 - A/]):]

== LEO cos ( 1 AOJ - : AS) cos ( o o t - Po:). (8 -51 )

Since Aw << wO, the exprwion in Eq. (8-51) represents a rapidly oscillating wave having :ln i~ngi~lar lrcquc~.cy in , , and a n i~mplitildc that varies slowly will] ;lo ;~npul;ir li-cquc~~cy All). 'I'his is depicted in Fix. 8- 5.

The wave inside theenvelope propagates with a phase velocity found by setting wot - p0z = Constant:

The velocity of the envelope (the group velocity 11,) can be determined by setting the argument of the first cosine factor in Eq. (8-51) equal to a constant:

t Aw - z AD = Constant, I

Fig. 8-5 Sum of two tihe-harmonic traveling waves of equal amplitude and slightly different frequenb~es at a given t.

324 PLANE ELECTROMAGNETIC WAVES / 8

from which we obtain 1 dz Aw u = - = -

dt Ap -ma In the limit that A o - 0, we have the formula for computing the group velocity in a dispersive medium.

This is the velocity of a point on the envelope of the wave packet. as shown in Fig. 8-5. and is identified as the velocity of the narrow-band signal. .,

A relation between the group and phase velocities may be obtained by combining Eqs. (8-50) and (8-52). From Eq. (8-50), we have

---. Substitution of the above in Eq. (8-52) yields

lip U g =

w du 1 --P up nw

From Eq. (8-53) we see three possible cases:

a) No dispersion :

du P = 0 (up independent of w, /3 linear function of w), dw

U g = U p .

b) Normal dispersion:

duP - -? 0 (up decreasing with w), . dw ll,, < 14,).

c) Anomalous dispersion:

dup - > 0 (up increasing with w), dw

Example 8-4 A narrow-band signal propagates in a lossy dielectric medium which has a loss tangent 0.2 at 550 (kHz), the carrier frequency of the signal. The dielectric

(8-52)

I

I Fig. 8-5, . .

ombining

lium which IC dielectric

I '

ent o/oc = 0.2 and 02/8(&c)' << 1, Eqs. (8-40) and (8-41) can a and ,O respectively. But first we find o from the loss tangent:

a = 1.53 x (S/m). Thus,

.- d

= 0.0182 x 1 .005 = 0.01 $3 (radlm).

b) I'hasc velocity:

From Eq. (8-41) we h ~ v e

c) Group velocity:

Since ug # 5, the medium is dispersive As we can see, the computed values of u, and up do not differ much because of the Small value of the loss tangent.

. % 8-4 FLOW OF ELECTROMAGNETIC i I - d

POWER AND THE POYNTING VECTOR i ' I I

Electromagnetic waves carry with them electromagnetic power. Energy is transported : through space to distant receiving points by electromagnetic waves. We will now 1 derive a relation between the rate ofsuch energy transfer and the electric and magnetic field intensities associated with a traveling electromagnetic wave.

We begin with the curl equations I

The verification of the following identity of vector operations (see Problem P. 2-23) [ is straightforward: I

I 1

V . (I3 x 1-1) = 1 1 . (V x E) - E . ( V x 1-11. (8 -56) .?

Substitution of Eqs. (8-54) and (5-55) in Eq. (5-56) yields -'-. In a simple medium, whose constitutive parameters r, 11, and D do not chan, oe with time, we have 1

dB H.- - - H . = - d(pH) l i ( p H s H ) - < ( l - ) d r at 2 i jt st p H 2

GO I I I nccrt: - 1s) (7 15 .---= E .---- - - (7 t (7t 2 (11

E - J = E - ( c E ) = a E 2 . I Equation (8-57) can then be written as i*,

V\(E x H ) = -- (8-58) e .( -

which is a point-function relationship. An integral form of Eq. (8-58) is obtiined by integrating both sides over the volume of concern.

6 (E x H) ds = oE2 do, (8-59)

where the divergence theorem has been applied to convert the volume integral of V - (E x H) to the closed surface integral of (L x H).

i magnetic

9) (8-54)

h) (8-55)

31 I'. 2-23)

(8-56)

t p

57)

hmge with

(8 -58)

1s obtained

0

5') ) --.

integral of

-ansported : + :'will now +

!

the first and second t&ms on the right side of Eq. (8-59) of ,change of the energy stored, respectively, in the electric

with Eqs. (13-i46b) and (6-151c).] The last term is volume as a rk?sult of the flow ofconduction current electric ficld E. Hence we may interpret the right

side of Eq. (8-59) as the rate of decrease of the electric and magneuc energies stored, subtracted by the ohmic power dissipated as Heat in the volumc V. Tn order to be consismt kith thc ltlw of conscrva~ion ol' cllcrgy, this must equal the power (rate of energy) leaviilg the volume through its surface. Thus the quantity (E x H) is a vector representing the power flow per unit are'd. Define

Quantity :P is knoivn a s rhc Poyillitlq ~ i c p r . L o / . , which is a power dcnsity vector asso- ciatcd with an' elec~roma~nctic ficld. The assertion that the surhce integral of .P over a closed surface, as given by the left side of Eq. (8-59), equals the power leaving the enclosed volume is referred to as P O > . ) I ~ ~ J I ~ ' S theorem.

Equation (8-59) may be writren in another form

where

we = ~ E E ' = Electric energy density,

w,, = $pH2 = Magnetic energy density,

p, = OE' = J ~ / O 7 Ohmic power density.

In words, Eq. (8-61) states that the total power flowing ittto a closed surface at any instant equals the sum of the rates of increase of the stored electric and magnetic energies and the ohmic power dissipated within the enclosed volume.

Two points concerning the Poynting vector &re worthy of note. First, the power relations given in Eqs. (8-59) and (8-61) pcrtain to the total power flow across a closcd surface ohtaincd by Lhc surhcc integral of (E x 1-1). The definition of the Poynting vector in Eq. (8-60) as the power density vector at every point on the surface is an arbitrary, albeit useful, concept. Second, the Poynting vector :P is in a direction normal to betkEE and H.

Ifthc region ofconccrn is losslcss (cr = 0). rhcn the I:N tcrm in Eq. 18-61) vanishes. ~ ind the total power flowing into a closed surface is equal to the rate of increase of the.stored electric and magnetic energies in the enclosed volume. In a static situation, the first two terms on the right side of Eq. &61) vanish, and the total power Howing into a closed surface is equal to the ohmic power dissipated in the enclosed volume.


. ,

Example 8-5 ' ~ i n d the Poynting vector on the surface of a long, straight conducting wire (of radius b and conductivity v) that carries a direct current I. Verify Poynting's theorem.

Solution: Since we have a DC situation, the current in the wire is uniformly dis; tributed over its cross-sectional area. Let us assume that the axis of the wire coincides with the z axis. Figure 8-6 shows a segment of length t of the long wire. We have

and J I E = - = a - -

' a - anb' On the surface of the wire,

Thus the Poynting vector on the surface of the wire is -\.

I" P = E x H = ( a Z x a,)-----

2an2b3

- 1' - -ar----

2 4 h 3 '

which is directed everywhere into the wire surface.

Fig. 8-6 Illustrating ~oynting's 1 theorem (Example 8-5). ,

8-4.1 Ir Power DE

onducting 'oynting's .

~rmly dis- coincides We have

8-4 I FLOW OF EL~CTROMAGNETIC P ~ W E R ~ N D THE POYNTING VECTOR 329 , . \ 1 i : ' I . , '

, I

In order to verify ,Poy$ting3s theorem, ye &negate P over the wall of the wire Segment in Fig. 8-6. 'i - i ' ,

. a

P . a , d ~ . = ' (*,Sb3) - 2nb'

where the formulz for the iesistance of a straight wire in Eq. (5-13), R = / /US, has bcen used. The aboved-esult afFjnnr that the negdtive surface integral of the Poynting , vector is exactly equal to the 1'R ohmic power loss in the conducting wire. Hence Poynting's theorem is verified. I

8-4.1 Instantaneous and AVer~ge Power Densities

In dealing w ~ t h time-harrronlc electromagnetic waves. we have found it convenlcnt to use phasor notations. The instantaneous value of a quantity is then the real part of the product of the phdsor quantity and eJu' wheh cos ot is used as the reference. For example, for the phasor

E(:) = axEx(:) = a , ~ ~ e - ( " + ~ ~ ! ' ,

the instantaneous expressim is

For a uniform plane wave propagating in a lossy medium in the +: direction, the associated magnctic fleld illtensity phasor is

where 0, is the phase angle of the intrinsic impedance q = Iq(rjopi of the medium. The corresponding instantaneous expression for H(z) is

h 0 H(z, t) = %[H(z)ej?'] = a, - e-' cos (or - /?z - 0,). (8 -64b) I4

This proccdurc is pcrmissihle ils long as thc operations andlor thc equations involving the quantities with sinbsoidal time dependence are linrur. Erroneous results will be obtained if this procedure is applied to such nonlinear operations as a product of two sinusoidal quantities. (A Poynting vector, being the cross product of E and H, falls in this category.) The reason is that ,

. , . * * . . -. a . I I . 330 PLANE ELECTROMAGNETIC WAVES 1 8

1 . -

The instantaneous expression for th;'poynting vectoi,or p&w density vector is on the one hand, from Eqs. (8-63a) and (8-64a),

E: = a= - e-2az cos (wt - /jz) cos ( ~ t - /3z - O,,) I 14

,5Z

= a, 4 e-'" [COS U,, + cos (20t - 2pz - O,)]. ! 2141

(8-65)'

1 On the other hand, !

E; 9e[E(z) x ~(z)e j@'] = a= - e-'"' cos (wt - 2/32! - O,), lvl

which is obviously not the same i s the expression in Eq. (8-65). As far as the power transmitted by an electromagnetic wave is con'cerned, its

average value is a more significant quantity tli;~n its inst;~iit;~nco~is v:duc.. Fro111 Eq. (8 - 6 5 ) wc obtain l h c tinic-;~\wxgc I 'oynli l ig \ c c t o r , .Y',,,,!:),

1.

where T = 2n/w is the time period of the wave. The second term on the right side of

' Conslder two general complex vectors A and B. We know that

Wc(A) = +(A + A*) and &(B) = $(B + B*),

where the asterisk denotes "the complcx conjugate of." Thus,

9 ( ( A ) x :'Ac(B) = +(A + A*) x +(B + B*) = J[(A x B* + A* x 4%) + (A x B + A * x B*)] = :%<(A x B* + A x U). (8-66)

This relation holds also for dot products of vcctor functions and for products of two conlplex scalar functions. It is a strniglitforward ekercise to obtain the restlit in Eq. (8-65) by identifying the ccctors A and B in Eq. (8-66) with E(:)eJm' and H(z)eJw' respectively.

Equation (8 -67) is quite similar to the formula for computing the power dissip;~tcd in an impcdnncc % = I%lriO. w l l c ~ ~ ;I si~~usoitl:~l volti~gc I ) ( / ) = 6, cos rt11 appears ucross its tcrn1in;tls. Tllc i n h t : ~ ~ ~ t ; t t ~ c o ~ ~ s expression for the current i(/) through thc impcdancc is

vo i ( t ) = - cos (ot - 0,).

1-4 From the theory of AC circuits, we know that the average power dissipated in Z is

where cos 0; is the power factor of the load impedance. The cos 0, factor in Eq. (8-67) can be considered the power factor of the intrinsic impedance of the medium.

ncerncd, its due From

i

-67)'

J right side of

(8-66)

:ornplex scalar ng the vectors

d n ~mpedance instdntaneous

A { \ - -

i he considere?

. \ 8 ,

EL,. (8-65) is a cosind Sunuiun of a doub~c frJquency whose average is zero over a 1 fundamental periad. : , ;,

;I Using Eq. (8-66),\we q n express the instan)hneous Poynting vector in Eq. (8-65) as the real part of ~ H L sum-of two terms, instelld of the product of the real parts of !wo complex vectors.' ': I

= ! ; R I [ E ( i ) x H*(r) + E(;) x H(.z)e~"~']. (8 -65) The average power 8ensity, Pav(z), can be obtained by integrating P ( z , t ) over a fundamental period;;?. Since the average of the last (second-harmonic) term in Eq. (8-68) vanishes, We have

I

, 3

.YIV(:) = i ~ 9 ~ [E(z) X f-I*(:)]. In the general case, wk may not be dealing with a wave propagating in the z direction. We write

, = . E x 1 I ) (W/mZ), (S -69)

which is 3 general forhu1.l for computing the average power density in a propagating . wave.

Example 8-6 The far firld of a short vertical currem element I dl located at the origin of a spherical coordinate system in free space is

E(R, 0) = a,,E,(R, 0) = a, (b";Ld' - sin 0) e-~p' ( ~ / m ) and

&(R, 0) H(R, 0) = a,,, ----- - - ( s O ) e J (Aim), '70 21.R where i = 271/lj is the wa\.elength.

a) Write the expression for instantaneous Poynting vector.

b) Find thc total avcragc power :diiiice by the current element.

Solution

a) We note that ~ , / k , =: '7, = 12On (Q). The instantaneous Poynting vector is

9 ( Z a : 4 = *&xE[E(R, O ) P J ~ I x ~C[H(R, e)ejmfl

I 332 PLANE ELECTROMAGNETIC WAVES 1 8

- b) The average power density vector is, from Eq. (8-69),

which is seen to equal the time-average value of B(R, 8 ; r) given in the first equation of this solution. The total average power radiated is obtained by integrating .Ya,(R. 0) ovcr thc surface of thc sphere of radius R.

I dd , , Total Pa. = $ s Pa"@, 0) . ds = SozR S: [ISn (x) sin2 0 ] R2 sin 0 dO1dd

2

= 40a2 (:) l 2 (W),

where 1 is the amplitude (3 timcs thc cn'cctivc valuc) of the sinusoidal currcnt in dd.

8-5 NORMAL INCIDENCE AT A PLANE CONDUCTING BOUNDARY

Up to this point wc have discussed the' propagation of uniform plane waves in an unbounded homogcneous medium. In practice, wavcs oftcn propagate in bounded regions where several media with different constitutive parameters are present. Whcn a n clcctrotnagnctic wave traveling in onc medium impinges on another medium with a dilkrent intrinsic impedance, i t experiences a rellcction. I n Scct~ons 8-5 and 8 -6 wc examine the bchavior of a planc wave when it is incidcnt upon a plane conducting boundary. Wave behavior at an intcrfacc betwecn two cliclcctric media will be discussed in Sections 8-7 and 8-5.

For simplicity we shall assume that the incident wave (Ei, Hi) travels in a lossless medium (medium l:o, = 0) and that the boundary is an interface with a perfect conductor (medium 233, = co). Two cases will be considered: normal incidence and oblique incidence. In this section we study the field behavior of a uniform plane

,'- wave incident normally on a plane conducting boundary.

Consider the.situation in Fig. 8-7 where the incident wave travels in the + z direction, and the boundary surfacc is thc planc z = 0. The incidcnt electric and magnetic field intensity phasors are:

l<,(z) = J ' l l z (8 -70a)

where Eto is the magnitude of El at z = 0, and 8, and 11, are, respectively, the phase constant and the intrinsic impedan3 of medium 1. It is noted that the Poynting vector of incident waves, Pi(z) = E,(z) x Hi(z), is in the a, direction, which is the direction of energy propagation. The variable z is negative in medium 1.

)idal current

waves in an in bounded .ire present. on mother

I . In Sections idcnt upon a wo dielectric

s inJF? lossless ith a perfect ncidcnce and nirorm plane

:Is in the + z t electric and

ely, the phase the Poynting , which is the 1.

Insiclc medium 2 [a perfect conductor) both electric and magnetic fields vanish. E, = 0. 11, = 0; hcncb no wave is iransn~itted across the boundary into the z > 0 region. The incident wave is reflected, giving rise to a reflected wave (E,, H,). The reflected electric field intensity can be written as

E,(I) = a,Eroe+jl'l=, (S-71)

where the positive sign in the exponent signifies that the reflected wave travels in the - z direction, as tliscusscd in Section 8-2. The total electric field intensity in medium 1 is the sum of E and E,.

E,(z) = E,(z) + E,(z) = a,(Eloe-JP1' + EroefJP1z). (8-72)

Continuity of the tangenti 11 component of the E-field at the boundary r = 0 demands that

I;,(O) = a,(Elo + E,,) = E,(O) = 0,

which yields Ero = -Aio. Thus, Eq. (8-72) becomes

El(z) f a , ~ , ~ ( ~ - j P l ' - e+jP1z)

= -a, j2Ei0 sin /I,:. (8 -73,)

The msirctic field il.tensity H, of the reflected wave is related to Er by Eq. (8 -24).

.. * I

Combining H,(z) with Hi(z) in Eq. (8-70b), we obtain the total magnetic field inten- t

> I . . _ . , . . , . '' sity in medium 1 :

. "

k - Hl(z) = Hi(z) + H,(z) = a,2 - Eio cos P1z. (8 -73 b)

'I 1

It is clear from Eqs. (8-73a), (8-73b). and (8-69) that no average power is associated with the total electromagnetic wave in medium I, since El(z) and Hl(z) are in phase quadrature.

In order to examine the space-time behavior of the total field in medium 1. we first write the instantaneous expressions corresponding to the electric and magnetic field intensity phasors obtained in Eqs. (8-73a) and (8-73b):

E,(z, t) = .%'o[~,(z)ej"'] = a,2Eio sin P I? sin r ~ . (8-74a)

Eio H ~ ( z , t) = .%e[H,(:)eJm'] = a,2 - cos Plz cos wt. (8 -74b)

'11

Both El(i, t ) and Hl(z, t) possess zeros and maxima at fixed distances from the conducting boundary for all t, as foilows:

1 -'.

Zeros of El(,-, t)

i i. occur at /Il: A -nx,

O r = = -"z' Maxima of H ,(z, t) n = 0, 1.2,. . . Maxima of El(;, i) )

7i ;. occuratB,z= -(2n+ I)-, or:= -(2n+ I) - ,

2 4 Zeros of Hl(z, t ) n = 0 , 1 , 2 . . . .

The total wave in medium 1 is not a traveling wave. It is a standing wave, resulting from the superposition of two waves traveling in opposite directions. For a given t. both El and H1 vary sinusoidally with the distance measured from the boundar) plane. The standing waves of E l = a,E, and H, = a,H, are shown in Fig. 8-8 for several values of wt. Note the following three points: (1) E l vanishes on the conducting boundary (E,., = - Eio); (2) H , is a maximum on the conducting boundary (H., = Hio = Eio/vl); (3) the standing waves of El and H, are in time quadrature (90 phase dillercncc) and are shifted in spiicc b y a qu;iricr wavelength.

Example 8-7 A ):-polarized uniform plane wave.(Et, Hi) with a frequency 100 (MHz) propagates in air and impinges normally on a perfectly conducting plane at x = 0. Assuming the amplitude of Ei to be 6 (mV/m), write the phasor and instan-

. .- taneous expressions for: (a) E, and Hi of the incident wave; (b) Er and H, of the reflected wave; and (c) El and H, of the total wave in air. (d) Dcierminc tile location nearest to the conducting plane where El is'zero.

1 1 icld inten- , , , 1 + I . ; { , I - . I t

(8-73h) ' . 1

issociated i e in phase

ium 1, we magnetic I

I

(8 -744

i Y -74b)

Irom the

,--

I

I ) :. 4

7 -, . . .

. resulting a given t , boundary 2. 5-8 for ! the con- boundary uadrature

1 e n c n 0 : I . .I ~d Instan- I f , of the : I~~c'l~lclll

Fig. 8-8 Standing waves of ;i, = r,E, md H I = r,H, for several values o f o ~ .

Solution: At the giveh frequency 100 (MHz), '

a) For the incident w i v e ( ( 1 travel~ng wave):

i) Phasor expressions

E,(x) = ily6 x 10-3e-J2nxi3 (v/m), 1

H,(x) = - ax X El(.;) = a, - e-J2"x'3 Y 1 2n (A/m).

336 PLANE ELECTROMAGN"ETIC WAVES 1 8

b) For the rejected wave (a traveling wave): . ,.

i) Phasor expressions . .

Er(x) = ,a,6 x 1@3ej2n"/3 (v/m),

ii) Instantaneous expressions

Er(x, r) = Wc[E,(x)eja'] = - 5 6 x cos 2n x 108t + - s 3

Hr(x, t ) = a= - cos 271 x 108t + lo-' ( , . 2n

c) For the total wave (a standing wave):

i) Phasor expressions

ii) Instantaneous expressions

E,(x, t ) = ~%e[E,(x)e~"'] = a 9 2 x sin -x sin(2n x 10") (? ) H (x , t ) = a, - cos (277 x 108t) (A/m).

d) The electric field vanishes at the surface of thc conducting plane at x = 0. In medium 'I, the first null occurs at

8-6 OBLIQUE INCIDENCE AT A PLANE CONDUCTING BOUNDARY

Whcn a uniform plane wave is incidcnt on a plane conducting surfacc obliquely, the behavior of the reflected wave depends on the polarization of the incident wave. In order to be specific about the direction of Ei, we define a plane of incidence as the plane containing the vector indicating the direction of propagation of the incident wave and the normal to the boundary surface. Since an Ei polarized in an arbitrary direction can always be decomposed into two components-one perpendicular and

Medium I , I * -

' Fig. 8-9 Planc wave inc~dcnt (01 = 0) obliqacly 011 a plane conducting

2 4 0 boundary (perpendicular polarization)

the other parallel to the pime of incidence--we consider these two cases sepnmtelj. The general case is ostilined by superposing the results of the two component cas~s.

8-6.1 Perpendicular ~ o l a r i r a t i o n ~

In the wsc of perpendicuiilr p d u r i ~ u t i o n , Ei IS p&+endicular to the plane of incidence. as illustrated in Fig. 8-9. Noting that

ani = a, sin Oi + a, cos Oi, (8-753 where Oi is the ungle of inci~lence measured from the normal to the boundary surface, we obtain, using Eqs. (8-17) and (8-23),

-

' Also referred to as horizontal poiari:ation or E-polcrrlzatioh.

i -, I . * 338 PLANE ELECTROMAGNETIC WAVES / 8 . -

At the boundary surface, z = 0, the total electric field intensity must vanish. Thus,

In order for this relation to hold for all values of x, we must have Ero = - Eio and I

Or = Oi. The latter relation, asserting that the angle of reflection eqtlals the artgle of 3

irtcideilce, is referred to as Snell's law oj' reJectiort. Thus, Eq. (8-78) becomes

The corresponding Hr(x, z ) is

The total field is obtained by adding the incident and reflected fields. From Eqs. (8-76a) and (8-79a) we have

El(x, z) = E,(x, z) + E,(.Y, Z )

= a E ( e - ~ 8 ~ Z C ~ d e , - &PI: cos8 , J b ~ x s r n O , Y 10 )e -

= - a,j2Eto sin (P,z cos Ol)e-J"lX e l . (8-80a)

Adding the results in Eqs. (8-76b) and (8-79b), we get

E;0 I

M,(.K, z) = - 2 - [a, cos U i cos (/jl; cos Ui)e-JI1~X"nol '1 1

+ a,j sin 0, sin (p lz cos Oi)e-j51xsin o 1 1. 18-80b)

Equations (8-8Oa) and (8;80b) are rather complicated expressions, but we can make the following observations about the oblique incidence of a uniform plane wave with polarization on a plane conducting boundary:

1. In the direction ( z direction) normal to the boundary; El, and H I , maintain standing-wave patterns according to sin Plzz and cos filzz, respectively, where f i l= = P I cos Oi. No average power is propagated in this direction since El, and HI, are 90" out of time phasc.

2. In the dircction (.K direction) parnllcl to thc boundary, E , , and If,, arc in both time and spacc phasc and propagate with a phasc vclocity

(I) V ) u1 U l x =-=----=-.

PIX /3, sin Oi sin 8,

1 ... : . . ' I . /

': ' ,..':;:, : i '

st vanish. ..: . , , I : . 1 r , ..

b ., ,, ! 1 . .. * , , :, 1.. . , ,.' ' I , , : .

- E,, and unyle of

:s

(8-79a)

(8-79b)

:on1 f-".

( x -- aoa)

is-sob)

t \VC call t11 plane

7 x 1 infain >. \\~llcl.o

L,. apq in .

' 3. The propagating wave in the r direction is a nonunijom plane wave because its amplitude varies witIi'z.

I . Since E, = 0 for all x when sin (P,z cos OiJ = 0 or when

^ 2n fllz cas oi = - A - cos 0, = - 1117t,

)' 1 n z = 1 , 2 , 3 , . . . ,

a conducting plilte can be insertdd at

without changinh tht: field pattern that exists between the conducting plate and the conducting ~~~~~~~~~y at I = 0. A tiitn.queiac elearic (TE) IIYIC~ ( E l , = 0) will bounce back and forth between the conducting pinnes ilnd propagate in tne r direction. We hitVe, i I effect, a parallel-plate waveguide.

Example 8-8 A uniforrrl plilhe wavc ( E l , H,) of an angular Srequcncy w ir lncldent from air on a very ILrge. perfectly conducting wail a r an angle of incidence 0, with perpendicular polarihllun. Find (a) thc current induced on the wall surface, and (b) the time-average Poynting vector in medium 1.

a) The conditions df this problem are exactly those we have just discussed: hence we could use the kormulas directly. Let r = 0 be the plane representing the rurfxe of the perfectly cbnducting wall. and let Ei be polarized in the y direction. as was shown in Fig. 8-9. At z = 0, El(x, 0) = 0, and H,(r. 0) can be obtained from Lq. (8-80b):

Inside the perfectly conducting wall, both E2 and H, must vanish. There is then a discontinuity ih iho milgnktic field. Thc amount of discontinuity is equal to ilic surfacu currchl. Iqroin Lcl. (7-52b), wc have .- -

-The instantaneous expression for the surface current is

Ei0 J,(x, t ) = a, - cos 4 cos o 6 0 ~

(8-82)

It is this induced current on the wall surface that gives rise to the reflected wave in medium 1 and cancels the incident wave in the conducting wall.

b) The time-average Poynting vector in medium 1 is found by using Eqs. (8-80a) and (8-80b) in Eq. (8-69). Since El, and HI, are in time quadrature. Pa, will have a nonvanishing x component.

E 2 - - a,2-C sin Oi sin' /l,,r,

' / I

where /IlZ = /3, cos 0,. The time-average Poynting vector in medium 2 (apr fcc t conductor) is, of course, zero.

8-6.2 Parallel polarizationt -1

We now consider the case of E, lying in the plane of incidence while a uniform plane wave impinges obliquely on a perfectly conducting plane boundary, as depicted in Fig. 8-10. The unit vectors a,, and a,,, representing, respectively, the directions of propagation of the incident and reflected waves. remain the same as those given in

Reflected wave

Incident wave

Fig. 8-10 Plane wave incident obliquely on a plane conducting boundary (parallel polarization).

' Also referred to as tiertical polarizatiotl or H-polarizatidt~.

(8-82)

:tcd wave

;. (8-80a) *. .Pav will

(8 -83)

(a perfect

/?

\)I n- 11e Cplc'l,., 117

cctlolls of L' given in

m -

8-6 1 OBLlOUE !NCIDENCE AT A'PLANE CONDUCTING BOUNDARY 341 , . f"', ,I

1

i ? Eqs. (8-75) and ($-I!). Boih Ei and Er now have components in x and z directions, whereas Hi and H, hdve only a y component. w e have, for the incident wave,

t 1, i 1

Er(x, 2) = EIO(ax cos 0, -' a: sin Oi)e-jbl(x 'ln ' 1 + z c o s U f ) , (8 -84a)

H,(x, = ay 5 e - ~ \ l ~ ( x s t n o i + z tb\ 0 1 ) .

"t 1 (8 -83b)

The reflected wave (E,, H:) have the following phasor expressions:

E,(x, k) = Ero(a, cos 0, +'a, sin Or)e-jP1(xsln cos Or), (8-85a) ' :

H , ( ~ , i) = - B I E r o e - ~ ~ ~ ( x s l n 8 r - z c o s ~ r ~ . (8-8jb) YI 1

A1 thc surklcc ol '~l ic pclfcct conductur, : = O, ~ h c tanyentiti1 compolicn~ ( t l x .Y component) of the total electric field intcnsitymust vanish for a11 s, or E,,(,Y. 0) - 6,,(.u, 0) .= 0. From Eqs. (8.- 841) ant1 (8-.X~:I), wc 1i;ivc

whlch requires Ero = - f;,, and 0, = 0, . The total electric field intensity in medlum 1 is the sum of Eqs. (8+84id and (8-85a):

Adding Eqs. (8-84b) and (8-85b). we obtain the total magnetic field intensity in medium 1.

Hl(s. z ) = H,(s, z) + Hr(x, z)

El 0 = ay2 - cos (/jlz cos O , ) C - J D ~ ~ ' ~ ~ " 4 , (8-86b) YI 1

I

The interpretation oi' Eqs. (8-86n) and (8-86b) is similar to that of Eqs. (8-80a) and (8-SOW b r the erpendicular-pol:~rizi~tio~~ cclsc, cxcept that E,( \ - . z), ~nbtc;id of fl,(s. :), now 11;)s 1x1 P 1 1 ; I I I Y :rnti : I : componc~it. Wc C O I I C ~ L I ~ C . I~ICI.L'~OI.C:

,1. In the direction (2 direction) normal to the boundary, E l , and H, , maintain standing-wave patte-ns according to,sin P1,z and cos Dl,z, respectively, where j,, = Dl cos Oi. No average power is-propagated in this direction, since El, and H,, are 90" out of t ipe phase.

. . , ..r, l I . < .


2. In the x direction parallel to the boundary, El: and H,, are in both time and space phase and propagate with a phase velocity u,, = u,/sin O,, which is the same as that in the perpendicular polarization.

3. The propagating wave in the x direction is a nonuniform plane wave.

4. The insertion of a conducting plate at z = -nzi , /2 cos Oi (nt = 1, 2, 3, . . .) where E ,, = 0 for all x will not affect the field pattern that exists between the conducting

* plate and the conducting boundary at z = 0, which form a parallel-plate waveguide. A transverse magnetic (TM) wave ( H ,, = 0) will propagate in the x direction.

8-7 NORMAL INCIDENCE AT A PLANE DIELECTRIC BOUNDARY

When an electromagnetic wave;s incident on the surface of a dielectric medium that has an intrinsic impedance different from that of the medium in which'the wave is originated, part of thc incidcnt powcr is rcflcctcd and part is transmitted. We may think of the situation as being like an impedance mismatch in circuits. The case of wave incidence on a perfectly conducting boundary disc&ed- in the two previous scctions is like terminating a generator that has a certain internal impedance with a short circuit: no powcr is transn~ittcJ i~ito thc conducting rcgion.

As before. we will consider separately, the two cases of the normal incidence and the oblique inciclencc of a unili,rrn planc wavc on a planc diclcctrii: medium. 130th media are assumed to be dissipationless (a, = o2 = 0). We will discuss thc wave behavior for normal incidence in this section. The case of oblique incidence will be taken up in Section 8-8.

Consider the situation in Fig. 8-11 where the incident wave travels in the + z direction and the boundary surface is the plane z = 0. The incident electric and

Transmitted Reflected +-

wave H, b":+ wave

* H t an1

Incident wave H,

Medium 1 Medium 2 (€2, P2)

z = o

Fig. 8-11 Plane wave incident normally on a plane dielectric boundary.

. . .) where onducting . late wave- . direction.

:n the +: ctric and

' Y 1

magnetic field intensity phasors are [ . . " 171

.< 2 I. ..,. ;".; k . . ,+. .; 4 1; , 3 . . - J , Ei(z) = a , ~ ~ ~ e - ~ p ~ , i : (8-873) Ei, H,(z) = a - e-JPlz.

I I (8-87b) 1 : t y ' I 1 3

These are the same expressions as those given in Eqs. (8-70a) and (S70b) . Note that z is negative in medltlm 1.

Because ofthe medium discontinuity at z = 0, the incident wave is partly reflected back into m e d i u ~ and partly transmitted into medium 2. We have

a) For the reJecte3 wave (E,, H,):

where El, is the magnitude of E, at r = 0, and p2'and '1, are the phase constant and the intrinsid impedance of medium 2 respectively.

Note that the directions of the arrows for E, and E, in Fig. 8-1 1 are arbitrarily drawn, because ErO and E,d may, themseives, be positive or negative. depending on the relative magnitudes pf the constitutive parameters of the two media.

Two equations $re needed for determinidg the two unknown magn~tudes Er, and El,. These equations are supplied by the boundary conditions that must be satisfied by the electric and magnetic fields. At the dielectric interface r = 0. :he tangential componed~s (the x components) of the electric and magnetic field intensltles must be continuous. b e have

~ ~ ( b ) + E.(O) = E,(O) or El, + Ero = El, (8-90a) and

1 El0 H.(o) + H.(o) = &o) or - ( E , ~ - E ~ ~ ) = -. 'I I

(8 -9Ob) -- 'I2

Solving ~ q s . &90n) nnd (8-90b), wo ihl;lio . 3

344 PLANE ELECTROMAGNETIC WAVES / 8 . .. I t

i - The ratios Ero/Eio and Eto/Eio are called, respectively, rejection coefficient and i

i transmission coefficient. In terms of the intrinsic impedances, they are

and L

El(z) = a,Eio[~e-jblz + T(j2 sin /I,z)]. (8 - 96)

We see in Eq. (8-96) that E1(z) is composed of two parts: a traveling wave with an amplitude rEio, and a standing wave with an amplitude 21.Ei,. Because of the existence of the traveling wave, E,(z) does not go to zero at fixed distances from the interface; it merely has locations of maximum and minimum values.

The locations of maximum and minimum jE,(z)l are conveniently found by rewriting E,(z) as

El(z) = a , ~ ~ ~ e - j " ~ ' ( l + TeJ2P1z). (8 -97)

t=-=- EtO *" (Dimensionless). EiO q2+q1

Note that the reflection coefficient I- in Eq. (8-93) can be positive or negative, depending on whether q , is greater or less than q,. The transmission coeflicient t, however, is always positive. The definitions for I. and t in Eqs. (8-93) and (8-94) apply even wh'en the media are dissipative: that is. even when 1 7 , and/or r 1 2 are corn- plcs. Thus I- ;rnd T may Ilic~nscl~cs hc co~nplcs in \lie y m x i 1 . c;lsc, A complcs (or r ) simply nicans tI1:11 :I pl~:~sc sllifl is i~i\roduccJ :I[ tile i ~ ~ t c r h c c ripuli 1.cllcctio11 (or tr:~nsrnission). RcI1c~li01i a ~ i d ~ ~ ~ ; i ~ i s n i i s s i ~ ~ ~ i c ~ ~ l l i ~ i c n t s ;IIX rclittcd by the following equation:

b i *

(8-94) 1

1 + T = T (Dimensionless). (8 -95)

If medium 2 is a perfect conductor, = 0, Eqs. (5-93) and (5-94) yield r = - 1 and t = 0. Consequently, Ero = - Eio, and E,, = O.'The incident wave will be totally reflected, and a standing wave will be produced in medium 1. The standing wave will have zcro and maximum points, as discussed in Scction 8-5.

If medium 2 is not a perfect conductor, partial reflection will result. The total electric field in medium 1 can be written as

El(,-) = Ei(:) + Er(:) = a , ~ ~ ~ ( e - j ~ ~ ' + rejal ' 1 = axEiO[(l + r ) e - J p l z + r ( e j p l i - ,-jPl: 11

. = a,Eio[(l + T)e-jpl' + T(j2 sin biz)] or, in view of Eq. (8-95),

8-7 I NORMAL INCIDENCE AT A PLANE DIELECTRIC BOUNDARY 345 I ' I

a i c i m and , : : i ! '1 For dissipationless media, 4 , and 4, are real, making both r and r also real. However,

I

:, I' can be positive or hegqtive. Consider the following two cases.

, , The maximum value of IE,(z)1 is Eio(l + n, which occurs when 2P,1,., = -2nn(n =0, 1,2,. . . ) , o r a t

The minimum 'value of lE,(z)l is Eio(l - r ) , which occurs when 2/I,zm,, = -(2n + l)n, or i t

The maximum vdue of /K , (z ) / is L..,(l - I-), wliicll occurs at .-,,,,, given in

Eq. (8-99); and the minimum valus of /E , (z ) is Eio(l + r), which occurs at

z,,, given in Eq,(S-OS). In other words. the locations for E,(;)I,,, and i i , ( ~ j l , , , ~ ~ when I- > 0 and when r < 0 are interchaflged.

The ratio of the m;iimum value to the minimum vdue of the ~ I C C ~ ~ I C fidd intensity of a standing wave is called the standing-wuae rutio, S.

I

An inverse relation of Eq. (8-100) is

S - 1 irl = s+r (Dimensionless). (8-101)

WIiib the v:ilua o f r rangx from - I :o i- i, the value oss riinges fro~n I to z. 1t IS

customary to express S on a logaritl 7;~. . i ' ~e . The standing-wave ratio in decibels 1s 20 log,, S. Thus, S = 2 corresponds io i) rt.ioalng-wave ratio of 20 log,, 2 = 6.02 dB and 1l-1 = (2 - 1)/(2 + 1) = ;. A. stnndiq-inve ratio of 2 dB is equivalent to S = 1.26 and (TI = 0.1 15.

4 ,

The rnapetic field intensity in mcdium 1 is obtained by combining Hi(:) and H.(z) in Eqs. (8'-87b) and (8-88b), respectively:


This should be compared with E,(z) in Eq. (8-97). In a dissipationless medium, r is real; and IH,(z)J will be a minimum at locations where /E,(z))I is a maximum, and vice versa.

In medium 2, (E,, H,) constitute the transmitted wave propagating in + z direction. From Eqs. (8-89a) and (8-94), we have

Et(z) = a , ~ E ~ ~ e - j P ~ ~ . (8-103a)

And from Eqs. (8-89b) and (8-94), t

Z H,(z) = a, - Eioe-jP2'. (8-103b)

72

Example 8-9 A uniform plane wave in a lossless medium with intrinsic impedance v l is incident normally onto another lossless medium with intrinsic impedance q, through a plane boundary. Obtain the expressions for the time-avcragc power dcnsirics in both ~nctlia.

.Pa, = $.%(E x H*) .

In medium 1. we use Eqs. (8-97) and (8-102).

where r is a real number because both media are lossless. In medium 2, we use Eqs. (8-103a) and (8-103b) to obtain

Since we are dealing with lossless media, the power flow in medium 1 must equal that in medium 2; that is,

h m , r- is num, and

f z direc-

(8-103a)

(8 - 103b)

npedance euance q , LAC power

12-~vcragc f i

(8 - 104)

csy-as, 4

m 1 must

J

Tha t Eq. (8-106) is true can be readily verified by using Eqs. (8-93) and (8-94).

3 ,a

8-8 NORMAL INCIDENCE AT MULTIPLE DIELECT R l ~ , l d f ERFACES

' f

In certain pi-@tical situations a wave m q be incident on severai layers of dielectric. media with difierznt constitutive parameters. One such situlition is the ~ 1 s t of a dielectric cbhting dn g!ass in order to reduce glare from sunlight. Another is n raclomc, which is il clomc-shapcd crlclosurc clcsigncd no1 only to protect radar installations from inclement weather but to permit the propagation of electromagnetic waves through the enclosure with as little refection as possible. In both siltla- tions. determining the propcl- dielcctric material and its thickness is an important design problem.

We now consider the three-region situation depicted in Fig. 8 - 12. A uniform plane wave traveling in the +: direction in medium.1 ( E , , p , ) impinges normally :it

a plane boundary with medium 2 ( E ? , p2) , a t : = 0. Medium 3 has a finite thickness and interfaces with medium 3 (c3, p 3 ) at z = d. Reflection occurs a t both : = 0 and z = d. Assuming ah .u-polarized incident field, the total electric field intensity in medium 1 can always be written as the sum of the incident component a,EiOe-JP'' and

wave +- H,

Medium 1 kll PI)

Hl Transmitted

h,lediunl 3 ( ~ h 13)

- 7 Fig. 8-12 Normal incidence 7 -

2 = 0 . . i = d at multiple dielectric interhces.

' , .,.. .

. . . . , k 1

,, . , . , . . , i . : ' i , , , , . .,::

. , . . . . . . . % . . 5 I

, ~, . . . , . , a reflected component ' a X ~ , , e j P i z : , . i : . ! : . , , -:-:.,:: ?,k.- >* . , , . . , I . . . . ,... . . . * . . !p:>:

1

However, owing to the existence of a second discontinuity at z'= d, Ero is no longer I

related to Eio by Eq. (8-91) or Eq. (8-93). Within medium 2 parts ofhaves bounce i

back and forth between the two bounding surfaces, some penetrating into media 1 and 3. The reflected field in medium 1 is the sum of (a) the field reflected from the interface at z = 0 as the incident wave impinges on it; (b) the field transmitted back )

into medium 1 from medium 2 after a first reflection from the interface at z = d; L i

(c) the field transmitted back into medium 1 from medium 2 after a second reflection ! * at z = d; and so on. The total reflected wave is, in fact, the resultant of the initial 8-8.1

reflected component and an infinite sequence of multiply reflected contributions within medium 2 that are transmitfed back into medium 1. Since all of the contributions propagate in the - z direction in medium 1 and contain the propagation factor ejhZ. they can be combined into a single term with a coefficient Ero. But hdw do we determine the rclation bctwocn I : , , , ;ind E,, , n o w ?

One way to find E,,, is to wrik don11 tilc clcctric ;tnd ~n:!gtictic licld illrct~sity vectors in all three regions and apply the boundary conditions. The 11, in region 1 that corresponds to the El in Eq. (8-107a) is, from Eqs. (8-S7b) and (8-SSb),

The electric and magnetic fields in region 2 can also be represented by combinations of forward and backward waves:

In region 3 only a forward wave traveling in + s direction exists. Thus.

1 On the right side of Eqs. (8-107a) through (8-109b), there are a total of four

unknown amplitudes: E,,, E l , E;, and E l . They can be determined by solving the four boundary-condition equations rccluirctl hy t h i continuity of ~ h c t:~ngcrlti:tl

I components of the electric and magnetic fields.

no longer I

es bounce I

3 media 1 I

1 from the itted back : at z = d ; l reflection &the initial ~trihutions : contribu- ;Ion factor low do we

ti intensity In region 1 4b). - m binations

31aI of four solving the : t a f i t i a l

. .

(8 - 1 10a)

. (8-110b)

8-8 1 ~ J O ~ M A L ~NCIDENCE AT' M U L ~ P L E DIELECTRIC INTERFACES 349 ! 4 ; L

(8-1 10d) 1 I

The procedure is stfaigl~tforward and is purely algcbraic (scc Problem P.8-23). In the following subsections we introduce the Concept of wave impedance and use it in an alternative approach for studying the prdblem of multiple reflections at normal incidence.

8 : I /. , /

. >

8-8.1 Wave Impedance 61 Total Field

We definc the wuur iiupedat~ce u j the ford jiuid at any piiine parallel to the plane. boundary as the ratio of the total electric field intensity to the total magnetic field intensity. With a I-dependent uniform plane wave as was shown in Fig. 8-12, we write, in general,

TotalEJz) '

Z(:) = (Q) . is-111, Total H,.(z)

For a single wave propapatinq in the + r direction in an unbounded mealum, the wave impedance eqilals the intrinsic impedance, q, of the medium; for a single wave traveling in the - z direction, it is - q for all z.

In the case of a uniform plane wave inddent fiom medium 1 normally on a plane boundary with an infinite medium 2, shch as that illustrated in Fig. 8-1 1 and discussed in Section 8-7. the magnitudes of the total electric and magnetic field intensities in medium 1 are, from Eqs. (8-97) and (8-102),

Their ratio defines the wave impedance of the total field in medium 1 at a distance : from the boundary plane

which is obviously d functiorl of.:. At a distance 2 = - L to the left of the boundary plane, .

1

Using the definition of r = (s, - q l ) / ( i l z + q , ) in Eq. (8-1 13), we obtain

'I2 c i s p l t + jq, sin P,L Z1(-.4 = 111

q l cos Pld + jq2 ~ i n . / 3 ~ t '

350 PLANE ELECTROMAGNETIC WAVES 1 8 .'

which correctly reduces to q, when q, = 7,. In that case, there is no discontinuity at z = 0; hence there is no reflected 'wave and the total-field wave impcdance is the same as the intrinsic impedance of the medium.

When we study transmission lines in the next chapter, we will find that Eqs. (8-1 13) and (8-1 14) are similar to the formulas for the input impedance of a trans- 1 mission line of length l that has a characteristic impedance q1 and terminates in an L

impedance q 2 There is a close similarity between the behavior of the propagation t . of uniform plane waves at normal incidence and the behavior of transmission lines. I

If the plane boundary is perfectly conducting, q2 = 0 and l- = -1. and Eq. f I

(8 -1 14) becomes ' i

Zl(-P) = jql tan P1d. (8-115)

which is the same as the input impedance of a transmission line of length C that has a characteristic impedance 1 1 , and ferminates in a short circuit.

8-8.2 Impedance Transformation with Multiple Dielectrics -1

The conccpl of lotal-liclci wavc impcdancc is \tc~.y useful in solving p r ~ b l e ~ ~ ~ s w t l l

multiple dielectric interfaces such as the shuation shown in Fig. 8-12. The total field in medium 2 is the result of multiple reflections of the two boundary planes 2 = 0 and z = d; but it can be grouped into a wave traveling in the + z direction and another traveling in the - 2 direction. The wave impedance of the total field in medium 2 at the left-hand interface z = 0 can be found from the right side of Eq. (5-1 14) by replacing y, by q , by q,, f i , by P,, and C by d. Thus,

q 3 cos P,d + jq, sin P,d Z2(o' = " q2 cos /12d + jq3 sin BJ'

As far as thc wavc in mcdium 1 is conccrncd, it cncountcrs o discontinuity at z = 0 and the discontinuity can bc charactcrizcd by an infinite medium with an intrinsic impedance Z2(0) as given in Eq. (8-1 16). The effective reflection coeficient at z = 0 for the incident wave in medium 1 is

. We note that I-6 differs from r only in that g2 has been replaced by Z2(0). Hence 1

the insertion of a dielectric layer of thickness d and intrinsic impedance q, in front - of medium 3, which has intrinsic impedancc y3. has the effect of transforming q , to Z2(0). Given q1 and q3, ro can be adjusted by suitable choices of q 2 and d. --

I I

. : - , Once I-, has been found from Eq. (8-117), Ero of the reflected wave in medium c

1 can be calculated: E,, = TOEi,. In many applications T o and E,, are the only 1 quantities of interest; hence this impedance-transformation approach is conceptudly

simple and yields the dcsired answers in :i dircct manner. It' the ficlds E j , L:; and <

b

f

. + I e -

:hat Eqs. f a trans- '

3 .

tcs in an '

paga4ion sn llnes. and Eq.

that has

,mt, 11

the only ceptually j

, E; and

$ _ I " I E, in media 2 and are also desired, t h y &n be determined from the bqundary conditions at z = 0 and z,= d (see Problem P,8-23).

71 i

Example 8-10 A dielectric layer of thickness d and intrinsic impedance q , is belwccn media 1 a n t 3 having intrinsic impedances q , and q3 respectively. Determine d and q2 su& that no reflection occurs wheh a uniform plane wave in medium 1 impinges normally on the interface with medium 2.

Solution: with the dielectric layer interposeti between media 1 and 3 as shown in . Fig. 8-12, thC condition of no I'eflection at intprface z = 0 requires To = 0, or Z2(0) =

q From Eq. (8:11~) wc havc i -

Equating the real and imaginary parts separately, we require

and

Equation (8-119) is satisfied if either

cos /?,d q 0 , which implies that

( )(I llld trlli' Ilitli~I, 1 1 ~ol l ( l l l iol l ( X 121) holdh, I:q. (8 120) C ; I I ~ bc s111sIicd when cilhcr (a) r l Z = t13 7 tll, ~ l ~ i c l i ib L ~ C Lrivi;11 case of 110 dis~~ntlnul t ies 21 ;ill, or (b) sin P,d = 0, or d = hA,/-.

On the other hdhd, il'relation (8-122) or (8-122a) holds, sin P,d does not vanish, and Eq. (8-120) cart be satisficd when 17, = 6. We have then two possibilities for thc condition of no rcllcction.

1. When q , = qi , we require 1- --. 2 2 d = i ~ - - ? 11 = 0, 1 , 2.. . .

2. d '

that is, the tqicknesr of the dielectric layer be a multiple of a half wavelength 6

' in the dielectric at (he operating frequency. Such a dielectric layer is referred to as a half-wave dielectric window. Since 1, = u,Jf = l / f a z , where / is the operating fre~ut!ncy. a half-wave diefectric window is a narrow-band device.

and 1 1 2 d = ( 2 n + I)-, n = 0 , 1 , 2 , . . . 4

When media 1 and 3 are different, q , should be the geometric mean of q , and 1 1 3 , and d should be an odd multiple of a quarter wavelength in the dielectric layer at the operating frequency in order to eliminate reflection 'Under these conditions the dielectric layer (medium 2) acts like a quarter-wave impedance tmnsformer. We will refer to this term again when we study analogous transmission-line problems in Chapter 9.

8-9 OBLIQUE INCIDENCE AT A PLANE n'K!.YCTR1C BOUNDARY

7 %

i b 2 now consider the case of a plane wavc that is incident obliquely at 311 arbitrary angle of Incidence Oi on a plane interface between two die1eC'ti-i.~-media. The medla are assumed to be lossless and to have different constitutive parameters (el, p,) and ( E ~ , p2), as indicated In Fig. 8-13. Bccausc of the medium's discontinuity at the interface, a part of the incident wave is reflected and a part is transmitted. Lines AO, O'A', and O'B are, respectively, the interscctions of the wavefronts (surfaces of constant phase) of the incident, retqected, and transmitted waves with the plane of incidence. Slnce both the incident and the reflected waves propagate In medium 1 with the same phase velocity up,, the distances m' and m' must be equal. Thus,

Reflected

Incident wave

Medium 1 ( € 1 . PI)

- 00' sin 0, = 00' sin Bi

Medium 2 ** Fig. 8-13 Uniform plane wave (€2, PZ) incident obliquely on a plane

z = 0 dielectric bo;ndary.

of 1 1 , and diekctric ~der these, n~pedance )us trans-

ardltrary h? I y - i l l a

, . I : , , .Ad :ty C

~:d . I , lncj ~ d x e s of : plme of ucd~um 1 . 'Thus,

f-'

I

-

,

I I

8-9 1 OBLIQUE INCIDENCE AT /# PPUNE DIELECTRIC BOUNOARY 353 1 . '. i

f : 4 1

f , t or . - . ,.:I r I > ! ., . . ' 5 ' ,

5 1

, 1 , 1-1 0, = [Ii. (8-123) -. i,

Eqaation (8 -123) asrllrcr or that tllc ;~nglc ok rcllcctioli is ~ ~ l ~ i i l to the .inglc of inc~dencc, which is S n e l l ' ~ luw oJreJecrion.

In medium 2, the timc it takes for the transmitted wave to travel from 0 to B equals the time for the incident wave to travel &om A to 0'. We have

.: , i aB AT . -- -- UP2 ~ P I

(Sii 0 3 ' ~ i n i l , I , , ~ ~ . - - -- X I ~ ~ , - s ~ O , -

1% l from which we obtain

sin 0 (8 - i Xi)

Furthermore, if medium reduces to

1 is free space such that 5, = 1 and n , = 1, Eq. (8-l?ib)

sin 0, 1 1

8-9.1 Total Reflection

Let us now examine !hell's law in Eq. (8-124b) for c, > e2-that is, when the wave in medium 1 is idcident on a less dense medium 2. In that case, 8, > Bi. Since 8, increases with Oi, an interesting 'situation arises when 0, = 7~12, at which angle


the refracted wave will glaze along the interface; further increase in Oi would result in no refracted wave, and the incident wave is then said to be totally reflected. The

a , I angle of incidence 0, (which corresponds to the threshold of total rejection 0, = n/2) , . -

is called the critical angle. We have, by setting 9, = n/2 in Eq. (8-124b),

This situation is illustrated in Fig. 8-14 whcrc a,,,, a,,,, and a,,, are uni t vcctors dc- ,noting the directions of propagation of the incident. reflected. and transmitted waves respectively.

What happens mathematically if Oi is larger than the-critical angle 0, (sin 0, > sin 8, = \iIEZ/E1)? From Eq. (8-124b) we have 1.

7

sin 8, = - sin 8, > 1, J : : (8-126)

which does not yield a real solution for 0,. Although sin 9, in Eq. (8-126) is still real. cos 8, becomes imaginary when sin 0, > 1.

Reflected wave

a,, Surface *

. "

z ... .

Incident , wave

Medium 1 ('I 1 KO)

,

, ~ e d i u m 2 * Fig. 8;14 Plane wave incident at J52: elqs ,-&

z = 0 critical angle, e l > e2.

ected. The n 0, = 74)

(8-125a)

(8-125b)

ktmors de- itted waves

0' (411 0 , > r

(b--126)

I is 51111 rcal,

(8 - 127)

/"

' 1

J ! A . , -

In medium 2, the unit vedtor n, in $he d k t i o d of propagation of a typical transmitted (refracted) waGe, as shown in Fig. 8-13, is

i - 1 4 r 1

a,, = a, sin 0, + a, cos 0,. (8-128) *J Both E, and H, vary .$atialIy in accordance with the following factor:

Tllc upper sign in hl. i H - 1.27) l l x bccn i~hmduned bi.c;;losc i t wo~ilcl ic;id to the impossihlc result of all inci-c;ising licld as 2 ii~crcasi.~. We can conclude frorn (8-119) that for Oi > 0, a wave exists o l m g the interface (in .Y direction), which is attenuated exponentially (rapidlyJ in nedium 2 in the,norm&l direction (i direction). This wave is tightly bound to the interface and is called a surface wove It is illustrated in Fig. 8-14. Obviously, it is li ncnuniform plane wave.

Example 8-11 A didlcctric rod or fibcr of*a .transparent material can be used to guide light or an electromagnetic wave under the conditions of total internal reflection. Determine the minimtim dielectric constant of the guiding medium so that a wave inc~dent on one end at any iing10 will bc confined wlthin the rod unt~l i t emerges from the other end. I!

J '

Solution: Refer to Fi . 8 -15. For total infernal reflection, 0, must be greater than or equal to 8, for the uiding dielectric medium; that is,

cos 0, 2 sin 8,.

Fig. 8-15 Dielectric rod or fibcr guiding elcctrornagnetic wave by total internal reflection.

356 PLANE ELECTROMAGNETIC WAVES 1 8 . . . A,.,

: r , . . . * . . 1 ,

1 ! .

t - ! From Snell's law of refraction, Eq. (8-124c), we have , .\ , I , . J I , ? ,

J c r ~ t

' , It is im~or t an t to note here that the dielectric medium has been designated as medium 1 (the denser medium) in order to be consistent with the notation of this subscction. Combining Eqs. (8-130), (8-131), and (8-125a), we obtain

Since the largest value of thk right side of (8-132) is reached when 0, = ni2, we require the dielectric constant of the guiding medium to be at least 2, which corre- <ponds to an index of refraction 1 1 , = 4. This requirement is satisfied by glass and

,

quartz. - --- 1.

We observe that Snell's law of refraction in Eq. (8-124b) and the crit~cal angle for total reflection in Eq. (8-125b) are independent of the polarization of the incident electric field. The formulas for the reflection and transmission coefficients, however. are polarization-dependent. In the following two subsections we discuss perpendicular polarization and parallel polarizatlon separately.

8-9.2 Perpendicular Polarization I

For perpendicular polarization the incident electric and magnetic field intensity : phasors in medium 1 are, from Eqs. (8-76a) and (8-76b): t

E ~ ( ~ , z) = a E , - j P ~ ( s s i n O , + : c o \ o t ) Y 10

(8-133a) i

The reflected electric and'magnetlc fields can be obtained from Eqs. (8-79a) and (8-79b), but remember that Ero is no longer equal to -El,. !

I E r b , z) = ayErOe - j / /~ (x 5rn 0, - : c o s 0,) (8-134a) 1

I n medium 2, the transmitted electric and magnetic field intensity phasors can - be similarly written as E

E , ( ~ , z) = a,,~,oe-jpz(.~ WIUI + CO* od (8-1351)

Et0 H,(x, I.) = - (-a, cos 8, '+ a; sin f3,)e-jfl"""'" et + zcO"t'. (8-135b) j . 'I2 I

i I

,, ,-> i F t

\ , .

(8-131)

ts medium ubsection.

intensity

-79a) and

i l

,. +; ' Therevare fokr anknown quantities in Eqs. (8-133a) through (8-135b), namely,

E,,, E,,, Or, ,hnd 8,. Their determlnarion follows from the requirements that the tangential compondnts of E and H be continuous at the boundary r = 0. From Bi,(x, 0) + E&, 0) .= E J x , O), we have

Because Eqs. (8-136a) and (8-li6b) are to he satisfied for ull x, all three exponential bctors that are function5 of x must be equal. Thus,

Plx sin Oi = /I,x sin 0, = /3,x sin Or,

which leads to Snell's law of reflection (8,. = 0;) and Snell's law of refraction (sin 0.1 sin @i = / 3 1 / P 2 = n1/h2!. Equations (8-136a) and (8-l36b) can now be written simply as

Eio + Ero = E,o and (8-137a)

1 ' 40 - (Eio - E,,) cos 0, = - cos Or, (8-137b) 'I1 '7 2

from which Ero and Et0 can be found in terms of Eio. We have

and

El0 TI=-= 2t12 cos Oi EiO '72 cos Oi + q cos Or

~ o r n ~ a r i n & ~ e expressions with the formulas for thc rellcction ;lnd transmission coefficients at normal incidsnce, Eqs. (8-93) and (8-94). we see th:~t the same formulas apply if ill and 17, are changed to (il,/cos O i ) and (rli/cos 0,) rcspcctively. When 0, = 0, making 0. = 0, = 0, these cxprcssions reduce to those for normal incidence, as they

358 PLANE ELECTROMAGNETIC.WAVES 1 8 -. i

should. Furthermore, T, and T, are related in the following way:

(8-140)

which is similar to Eq. (8-95) for normal incidence. If medium 2 is a perfect conductor. q , = 0. We have T, = - l(E,, = - Eio)

and T, = O(E,, = 0). The tangential E field on thc surface of the conductor vanishes, and no energy is transmitted across a perfectly conducting boundary, as wc have noted in Sections 8-5 and 8-6. 4

Noting that the numerator for the reflection coefficient in Eq. (8-138) is in the form of a difference of two terms, we inquire whether there is a combination of q , , q 2 , and €Ii, which makes T, = 0 for no reflection. Denoting this particular Oi by O,,, we require ,

q , cos OB1 = q 1 cos 0,. (8-141)

Using Snell's law of refraction, we have ,

cos 0, = J- = 1 - 5 siii- J (8-142) I7 5

and obtain from Eq. (8-141)

The angle 8,, is called the Brewster angle of no reflection for the case of perpendicular polarization. For nonmagnetic media, p , = p, = po, the right sidc of Eq. (8-143) becomes infinite, and 0,$, docs /lot exist. In thc casc of E , = c2 and p1 # p 2 , Eq . ( 8 - 1 43) reduces to

sin 0,, = 1

J = - G ~ '

which does have a solutioli whether p,/p, is greater or less than unity. However, it is a very rare situation in electromagnetics that two contiguous media have the same permittivity but different permeabilities.

Parallel Polarization

When a uniform plane wave with parallel polaiization is incident obliquely on a plane boundary, as illustrated in Fig. 8-1 6, thc incident and rcflectcd elcctric and magnetic field intensity phasors in medium 1 are, from Eqs. (8-84a) through (8-85b):

E~(x, 2) = ~ ~ ~ ( a ~ cos Oi - a= sin O i ) e - j p l ( x s i n O l + z c o s ' 3 ) (8-145a)

(8-140)

.O = - E d r vanishes, IS we have

5) is ill the .tion of Ill, 4 by 8,,?

(8-141)

(8-142)

/'-

(8-143)

pendicul~r q 18-143) $. 18- 143)

( 8 - 144)

Ho\+cver, i have the

juely on a ectrir(3 h ( 5 ):

(8-145a)

(8-135b)

I

E,(x, Z) = C,,(a, cos 0, + 3, sin ~ , ) L ' - j b ~ ( ~ s i n O r - = 0 s or) (s-1463) Er0 H,(x, 1) = - 3y -. - j D l ( * >;.I 0, - 2 a s 8,)

' I 1 (8-146b)

The transmitted electric and magnetic field intensity phasors in medium 2 are

Et(x, Z) = Eto(as cos 6, - a_ ,sin d,)e-jb:(xsinO~ + = m d t ) (8-1472)

z) ay 5 e - j , , ~ ~ x ~ i c 0, + z c w o t ,

'I2 (8-147b)

Continuity requirements for the tangential components of E and H at ; = 0 lead again to Snell's laws of reflection and refraction, as well as to the following two equations:

1 1 10 ' (8-148b) - (El, - E,,) = - E 'I I 11 2

Solving for Ero and E,, in lerms o[ELO, we obtain

Ero '12 cos 6, - cos Oi --= -1- I I - Eio q2 cos 8, + 'I, cos Oi

and

4 0 T I ! = - = 2112 cos 13,

Eio Y / ~ C O , S Q ~ + ~ ~ C O S O ~ '

' These are also referred to as Fresnel's e&ariorls.

360 PLANE EL~CTROMAGNETIC WAVES / 8

It is easy to verify that

Equation (8-151) is seen to be different from Eq. (8-140) forperpendicular polarization except when. Oi = 9, = 0, which is the- case-for normal incidence. At normal incidence rll and r l l reduce to r G e n in Eqs. (8-93) and (8-94) respectively, as did I?, and 7,.

If medium 2 is'a perfect conductor (qz = 0), Eqs. (8-149) and (8-150) simplify to rll = - 1 and r l l = 0 respectively, making the tangential component of the total E field on the surface of the conductor vanish. as expected.

From Eq. (8-149) we find thai TI , goes to zero when the angle of incidence Bi , , -

equals €IBl1, such that q 2 cos 9, = rl, cos OBtl

which, together with Eq. (5-142), requires

1 - ~ ' E ~ , / ' / L , € ~ sin' OBI1 = .

1 - .

The angle OBI1 is known as the Brewster angle of no reflection for the case of parallel polarization. A solution for Eq. (8-153) always exists for two contiguous nonmagnetic media. Thus, if p , = p2 = pO, a reflection-free condition is obtained when the angIe of incidence in medium 1 equals the Brewster angle OBI,, such that

Because of the difference in the formulas for Brewster angles for perpendicular and parallel polarizations, it is possible to separate these two types of polarization in an unpolarized wave. When an unpolarized wave such as random light is incident upon a boundary at the Brewster angle O,,,, givcn by Eq. (8-IS), only the component with pcrpcnclici~lar polarization will bc rcllcctctl. 'I'lii~s, a Iircwstcr anglc is also referred to as a polarizing utzyle. Based on this principle, quartz windows set at the

' Brewster angle at the ends of a laser tube are used to control the polarization of an emitted light beam.

Example 8-12 The dielectric constant of pure water is 80. (a) Determine the . , Brewster angle for parallel polarization, O,,,,, and the corresponding angle of trans-

mission. (b) A plane wave with perpendicular polarization is incident from air on water suiface at Bi = OBI1. Find the reflection and transmission coefficients.

REVIEW QUESTIONS 361

poldriza- .t normal pectivel y,

' I

) simplify ' the total

:idence 0,

'8-152)

r\ I ' J )

I( p~ ra l i d connag- when the

(8-154)

)endicular larization s incident Imponent .le is also het at the tion of an

/7

i - n ~ i ~ i e - - r l ~ c : of trans- )m air on

Solution i

a ) The Brewster angle of no reflection for darallel polarization can be obtained directly from Eq. (8-154):

I

- Osll = sin- ' ' 1 Jl+(ilE,2j

1 =sin- 'd . = 8 l . P .

1 + (1/80)

The corresponding angle of transmission is, from Eq. (8-124c),

1 ' = sin-' (;G) = 6.38".

b) For an incident Wave with perpendicu!ar polarization, we use Eqs. (8-138) and (8-139) to find TI m a r , at 0, = 81.0" and 0, = 6.38':

q1 = 377 (R), ql/cos 0, = 2410 (Rj 377

q z = - - - 0 . 1 ( ~ J C O S 0. = 40.4 (R).

Thus

We note that the h l a t ~ o n between TL and r , given in Eq. (8-140) is satisfied.

REVIEW QUESTIONS

R.8-1 Define uiliform plane Imve.

R.8-2 What is a wnvefionr?

R.8-3 Write'th4~ornogeneo.1~ vector Hclml~oiiz's equation for E in free s p c e .

N.8-4 Dclinc n r ~ n w ~ ~ n d r ~ l : t.hh is \v(.i~\cllt~t~lb~r icliiicd 10 ~ v a v c l c ~ ~ g ~ h ?

R.8-5 Define phase ve[dcity.

R.8-6 Define intrinsic itnpedance of a mediurn.'What is the value of the intrinsic impedance of free space?


R.8-7 What is a TEM wave? - 1 R.8-8 Write the phasor expressions for the electric and magnetic field intensity vectors of an x-polarized uniform plane wave propagating in the z direction.

R.8-9 What is meant by the polarization of a wave? When is a wave linearly polarized? Cir- 1 cularly polarized? 1 I R.8-10 Two orthogonal linearly polarized waves are combined. State the conditions under which the resultant will be (a) another lincarly polarized wave, (b) a circularly polari~ed wave, and (c) an elliptically polarized wave.

R.8-11 Define (a) propagation constant, (b) attenuation constant, and (c) phase constant. i R.8-12 What is meant by the skin depth of a conductor? How is it related to the attenuation ? constant? How does it depend on a'? p n J ? I

! R.8-13 What is meant by the dispersion of a signal? Give an example of a dispersive medium. I

t

R.8-14 Define group uelocit~. In what ways is group velocity different from phase velocity? , PROE

, - R.8-15 Define Poynting vector. What is the SI unit for this vector?-\

R.8-16 State Poynting's theorem.

R.8-17 For a time-harmonic electromagnetic 'field, write the expressions in terms of electric and magnetic field intensity vectors for (a) instantaneous Poynting vector and (b) time-average Poynting vector.

R.8-18 What is a standing wace?

R.8-19 What do we know about the magnitude of the tangential components of E and H at 1

the intcrfacc whcn a wavc impingcs normally on a pcrfcctly conducting planc boundary? i I

R.8-20 Define plane of incidence.

R.8-21 What do we mean when we say an incident wave has (a) perpendicular polarization and L:

(b) parallel polarization? i I

R.8-22 Define refiectiorl coejficient and rransritission cotfl'cier~t. What is the relationship between I

them?

R.8-23 Under what cohditions will reflection and transmission coefficients be real?

R.8-24 What are the values of the reflection and transmission coeficients a t an interface with a perfectly conducting boundary?

li.3-25 A pl:~nc w:~vc origin:~li~)g i t ) 111cdiu111 I ( r I , 1 1 , - / L , , . m l = 0) ix ~ I I C ~ C ~ C I I I ~ lot .~~l i~lfy OI I i t

plane intcrhcc ~ i l h medium 2 (6, # E,, p, = p,, a, = 0). under what condition will ~ h c electric field at the interface be a maximum? A minimum?

R.8-26 Define standing-wave ratio. What is its relationship with reflection coefficient?

R.8-27 What is meant by the wave impedance of the total field. When is this impedance equal to the intrinsic impedance of the medium?

'

ectors of an

Iar~/zd? Cir-

Lt~ons under 'inzed wave,

'ant.

: attenuation

w e medium.

elocity?

f-'

T J r i c t~ ,~ lc -~ \e rage

11 T. ,mcl 14 at I[! 1 1 ) !

aru.tlon and

1s111p between

7

merface with

ormally on a 111 ~ h e n t r i c

Y

cnt?

xdance equal

PROBLEMS 363

R.8-28 Thin dielectric cbating is sprayed on optical hstruments to reduce glare. What factors determine the thickness df the coating?

R.8-29 How should the thickness of the radome in a radar installation be chosen'?

lt,N-30 Si;~tc Stlcll's ltrw IJ/ ' w/lvc/iorl.

R.8-31 State Snell's law of rejklcriorl.

R.8-32 Define cr~rrcul dhyle. When does 11 ex~st at an interl'ice of two nonmagnetic rned~a"

R.8-33 Define Brewstet angle. Whcn does ~t exist at ah interface of two nonmagnetic media?

R.8-34 Why is a Brewster angle also called a polarizing angle?

R.8-35 Under what conditions will the reflection and transmission coefficients for perpendicular polarization be the same as those for parallel polarization?

PROBLEMS

P.8-1 Prove that thc electric field intensity in Eq. (5-17) satisfies the homogeneous Helmho1:z's equation provided that the condition in Eq. (8-18) is satisfied.

P.8-2 For a harmonic uniform pl:m wave propasating in a simple medium. both E and H '

\Iilry in accordance with thi hctor crp (-,jk R) as indicated in Eq, ( Y -?I 1. Show :ha: the forr S1:lxwcll's cquaiioi~s Ibl. unilorm pla~lc W;L\C in :I source-frcc rcgiun L . O ~ I I C O LO LI1c k~llowii l~:

k x I < ' = (I)/L!~I

k x H = --WEE k . F; = 0

k.11 = O .

P.8-3 The instantoncaus ixprcssioo for the magnetic field intensity of a uniform plane w:ivc propagating in the + y direction in air is givcn by

a) Determine k , alld tt-.e location whzre H z vanishes at r = 3 (ms). b) Write the instantaneous expression for E.

,,-- 1'3-4 Show that a plane u.avc with an instant;lncous expression for the electric field

E ( , 0 = a,EI0 sin (cot - k:) + askzo sin (rot - l i ~ + $)

is ellipticalIn~larized. Find the polarization ellipse.

PA-5 I ' ~ O V ~ tlic rollo\~i~ig

:I) An elliptically polarikd plane wave can be resolved into right-hand and left-hand circularly polatfzed waves.

b) A circularly polarized plane wave can bg obtained from a superposition of two oppositely directed elliptically polarized waves. .


I

P.8-6 Derive the following general expressions of the attenuation and phase constants for conducting media:

a = OJ &[/a - I]"' (Np/m)

P.8-7 Determine and compare the intrinsic impedance, attentuation constant (in both Np/m and dB/m), and skin depth of copper [a,, = 5.80 x lo7 (S/m)], silver [c,, = 6.15 x lo7 (S/m)]. and brass [c,, = 1.59 x lo7 (S/m)] at the following frequencies: (a) 60 (Hz), (b) 1 (MHz), and (c) 1 (GHz).

P.8-8 A 3 (GHz), y-polarized uniform plane wave propagates in the + x direction in a nonmagnetic medium having a dielectric constant 2.5 and a loss tangent lo-'.

a) Determine the distance over which the amplitude of the propagating wave will be cut ' in half. b) Dcterlninc the intrinsic impcd;lncc. thc w;~vclcngth, the p lme vclocity. and thc group

vclocity of the wave in the medium. '\.. c) Assuming E = a,50 sin (671 109t + n/3) at r = 0, write the instantaneous expression for

H for all r and x.

P.8-9 The magnetic field intensity of a linearly polarized uniform plane wave propagating in the + y directi~ii in sea water [E, = 80. p, = 1. c.= 4 (S/m)] is

'a) Determine the attenuation constant, the phase constant, the intrinsic impedance, the phase velocity, the wavelength, and the skin depth,

b) Find the location at. which the amplitude of H is 0.01 (A/m). c) Write the expressions for E(y, t) and H(p, r) at y = 0.5 (m! as functions o f t .

P.8-10 Given that the skin depth for graphite at 100 (MHz) is 0.16 (mm), determine (a) the 'conductivity of graphite, and (b) the distance that a 1 (GHz) wave travels in graphite such that its field intensity is reduced by 30 (dB).

P.8-11 Prove the following relations between group velocity u, and phase velocity up in a dispersive medium:

' P.8-12 There is a continuing discussion on radiation hazards to human health. The following calculations will provide a rough comparison.

a) The US, standard for personal safety in a microwave environment is that the power dcnsity be less than I0 (mW/cm2). Calculate thc corresponding stmdard in terms of electric field intensity. In terms of magnetic field intensity.

b) It is estimated that the earth receives radiant energy from the sun at a rate of about 1.3 (kW/m2) on a sunny day. Assi~ming a monochromatic planc wavc, calculate thc amplitudes of the electric and magnetic field intensity vectors in sunlight.

istants for .

0th Np/m 0' (Sin)], AHz), and

I

In a non-

dance, the

.ne (a) the such that

following n

thr;' - 7

\ ltn~,.- .d

: of about culate the

PROBLEMS 365

P.8-13 Show that the instantaneous Poynting vector of a propagating cikularly polarized plane wave is a constant that is indcpendent of time add distance.

P.8-14 Assuming that the radiatioh electric ficld interlsity of an antenna system 1s

E = a,& + a&,,, -

find the expression for the average outward power flow per unit area.

P.8-15 From the point of view of electromagnetics, the power transmitted by a IOSSIC:~ c~axial cable can be considered in terms of the Poynting vector inside the dielectric medium between the inner conductor and the otiter sheath. Assuming that a DC voltage V, applied between the inner conductor (of radius a) and the outer sheath (of inner radius b) causes a current I to flow to a load resistance, verif) that the integration of the Poynting vector over the cross-sectional area of the dielectric medium equals the power VJ that is transmitted to the load.

P.8-16 4 right-hand c~rcularly polarized plane wave represented by the phasor

impinges normally on a perfectly conducting wall at : = 0.

a) Determine the polarization of the rcllec;ted wave. b) Find the induced current on the conducting wall. c) Obtain the instdhtaneous expression of the total electric intensity based on a cosine

time reference.

P.8-17 A uniform sinusoidal planc wavc in air with the following phasor expression for electric intensity

is inciclcnt 011 a pcrkctly cond~~cting pl;~nc at : = 0.

:I) k'iritl thc I'rcclwncy :ml wavclcngll1 o f llic wave. b) Write the instantaneous expressions for E,(x, z ; t ) and Hi(x, z ; t ) , using a cosine reference. C) Determine the angle of incidence. d) Find E,(x, z ) and H,(?c, z ) of the reflected wave. e) Find E,(.u, :j and H ,(x, z) of the total field.

P.8-19 For the case of oblique incidence of a uniform plane wave with perpendicular polnriza- tion on a perfectly conducting plane bounhary as shown in Fig. 8-9, write (a) the instantaneous expressions

for the total fieldjn?edium 1, using a cosine reference; and (b) the time-average Poyctjng vector.

P.8-20 For the case of oblique incidence of a uniform plnne wave with p:wallel polarization on p~rkc t l !~ ond doc tiny pl;~nc Soul~clary as s l ~ u w ~ ~ in 1"ig. S-10. wilt: (a) lhc instantaneous

csprcssions

E,(x, z ; r ) and ' H,(s , z ; r)

for the total field in medium 1, tising a sine reference; and (b) the time-average Poynting vector.

366 PLANE ELECTROMAGNETIC WAVES 1 8 I * I :

P.8-21 Determine the condition under which the magnitude of the reflection coefficient equals that of the transmission coefficient for a uniform plane wave at normal incidence on an interface between two lossless dielectric media. What is the standing-wave ratio in dB under this condition?

P.8-22 A unifomi plane wave in air with E,(z) = a,10e-j6' is incident normally on an interface \ at : = 0 with a lossy medium having a dielectric constant 2.5 and a loss tangent 0.5. Flnd the following:

t a) The instantaneous expressions for E,(:, t), H,(z, t), E,(z, t) , and H,(z, t ) , using a cosine

reference. b) The expressions for time-average Poynting vectors in air and in the lossy medium.

P.8-23 Consider the situation of norn~al incidence at a lossless dielectric slab of thickness tl

in air. as shown in Fig. 8-12 with

e l = E , 7 E , and p, = p, = 11,. i a) Find E,,, E,', E;, and E,, in terms of E,,, d, 6,. and 11,. b) Will thcrc l x rcllcction nt intcrfircc : = 0 if tl = i L , 4 ' ? Espl:lin. I

P.8-24 A transparent dielectric coating is applied to glass ( E , = 4, p, = 1) to eliminate the reflection of red light [ I = 0.75 (pm)]. 1-

a) Determine the required dielectric constant and thickness of the coating. b) If violet light [ I = 0.42 (pm)] i s shone ormal mall^ on the coated glass, what percentage /

of the incident powcr will be reflected? I

P.8-25 Refer to Fig. 8-12, which depicts three different dielectric media with two parallel t i

interfaces. A uniform plane wave in medium 1 propagates in the + z direction. Let TI, and T13 t

denote, respectively, the reflection coefficients between media 1 and 2 and between media 2 and !

3. Express the effective reflection coefficient, I-,, at 2 = 0 for the incident wave in terms of T I : . r,,, and P,d.

i !

1

Medium 1 ( € 1 9 PI) Fig. 8-17 Plane wave incident

normally onto a dielectric slab backed by a perfectly conducting

z = O z = d plane (Problem P.8-26).

I

lent equals In interface I t condition?

In interfiicc 5 . Find the

~g a cosine

adium. ,

thickness d

;no parallel i - , : and rZ3 media 2 and srms of r,,,

f-?

. I ' . PROBLEMS 367

. . ( I

!'..'. , ' 1 ,

PA-26 A uniform wave with

- E,(i, t) = a,E,, cos CCI

I .

in medium 1 (el, p i ) is incident normally onto a lossless dielectric slab ( E , , p,) of a thickness d backed by a perfectly coriducting plane, as shown in Fig. 5-17. Find c.

a) E,(& t ) b Elk , 4 Ez(z, 0 4 P a V h 4 (9'A - f ) Determine the thickness d that makes E1(3, t ) the same as if the dielectric slab were absent. ;*

P.8-27 A uniform plane wave with E,(z) = a,~~,,e-jfl~' in air propagates normally through a

, thin copper sheet of thickness (1, as shown in Fig, 8-18. Neglecting multiple reflections within the coppcr shccts, lind

;I) E f . 111 1)) 1;: , l l j C) E 3 0 , H30 4 v'JA'(,pJL

Calcuiate (9aP,,)3/(9'av)i for a thickness d that equals one skin depth at 10 (MHz). (Note that this p~rtains to the shieldin$ elrcctivcncss of thc thin coppcr shcct.)

Copper sheer

t Z

(€0. POI - ,d+

* A . r

Fig. 8-18 Plane wave prop::gating through a t,liin copper sheet (Problem P.S-27).

P.8-28 A lCL(k-Hz) parullcll) polarized elcctromagnctic wave in air is incident obliquely on :In ocean surface at a near-grazlrlg angle Oi = SS'. Using E , = 81. jl, = I , :~nd a = 4 (SJln) for sea- wntcr, find (a) thc anglc of rcfi.nctian 0,. (b) thc transmission cocificient T ~ ~ . (c) (.Pd,,),:(.Pd,j,, and (d) the distance below tllc ocean surface where the field intensity has been diminished by 30 (dB).

P.8-29 A light ray is idcitienr from air obliquely on a transparent sheet of thickness d with an index of refraction n, as shown in Fig. 8-19, The angle of incidence is Oi. Find (a) 0,. (b) the distance L, at the point of exit, and (c) the amount of the lateral displacement l2 of the emerging ray.

Fig. 8-19 Light-ray impinging. , f .. - + ,

a + A obliquely on a transparent sheet of . . v ". - . .t' . refraction index n (Problem P.8-29).

P.8-30 A uniform plane wave with perpendicular polarization represented by Eqs. (8-133a) and (8-133b) is incident on a plmdinterface at .- = 0, as shown in Fig. 8-13. Assuming tr < r , and 0 , > Oc, (a) obtain the phasor expressions for the transmitted field (E,.?,), pnd (b) verify that the average power transmitted into medium 2 vanishes.

I\

p.8-31 Elcctromaglletic w;,vc from il~ldenv;~tcr source with perpcndlci ir p d ~ r i ~ l t i c ~ l l i.: P incident on a wrter-air interf~lcc at t i i = 20, , Using cr = 81 rlidp,= 1 fresh uatcr. lind

(a) critical angle 8.. (b) reflection coefficient r,, (c) ti;msmission coefficient 7,. and id) attenuation indB for each wavelength into the air. -. P.8-32 Glass isosceles triangular prisms shown in Fig. 8-20 are used in optical instruments. Assuming t, = 4 for glass. calculate the percentage of the incident light power reflected back by the prism.

7 . . Incident

1 *- ' - . .

Fig. 8-20 Light reflection by a right isosceles triangular prism (Problem P.8-32).

-, %. . -* r

P.8-33 Prove that, under the condition of no reRection a an interface. the sum of the B r e w e r 1: angle and the angle of refraction is n/2 for:

k a) perpendicular polarization ( p i # pz), . , i f . b) parallel polarization ( E ~ # c t ) .

, P.8-34 For an incident wave with parallel polarization:

t . .' . ~ . a) Find the relation between the critical angle 0, and the Brewster angle OBI, for nonmagnetic

i media. I b) Plot 0, and esll versus the ratio E , / E ~ .

1 L

I

P.8-35 By using sncll's jaw of refrxtion, (a) express r and r in tcrnms of r.,, f 2 , and 0 , ; and . t? (b) plot I- and T versus 0, for r,,/e,, = 2.25.

P.8-36 In some books the rcflection and transnmsion coerXcicnts for parallel polar~zatlon are defined as the tatios of the amplitude of the tangentla1 components of, respectively, the retlected and transmitted E fields to the amplitude of the tangential component of the incident E field. Let the coeficients defined in this manner be destgnated, respect~vely, Ti, and ril.

a) Find ril and Til in terms of j l l , t7:, O , , and 8,; and compare them wlth TI, and r , , in Eqs. (8- 149) alld (8-150).

b) Find the rslatioi between Ti, and ri,, and compare if with Eq. (8-151).

9-1 INTRODUCTION

We have now developed an e~ectrdma~netic model with which we can analyze electromagnetic actions that occur at a distance and are caused by time-varying charges and currcnts. These actions are explained in terms of clcctrorna~netic fields and waves. An isotropic or omnidirectional electromagnetic source radiates waves equally in all directions. Even when the source radiates through n hi$-ty directive antenna. its encrgy spreads over n widc.arca at Iargc dishnccs. This radiatctl cncrgy is n o t guided, and the transmission ol' power andinformation from thc source to a receivcr is incficient. This is cspccinlly true at lowcr frcqucncics for which dircctivc antcnnas would have huge dimensions and, therefore, would be excessively expensive. For instance, at AM broadcast frequencies, a single half-wavelength antenna (which is only mildly directivet) would be over a hundred meters long. At the 60-Hz power frcqucncy a wavelength is 5 million meters or 5 (Mm)!

For cllicicnt point-to-point transmission of power and information, thc source energy must be directed or guided. In this chapter we study transverse electromagnetic (TEM) waves guided by transmission lines. The TEiM mode of guided waves is one in which E and H are perpendicular to each other and both are transverse to the direction of propagation along the guiding line. We have discussed thc propagation

.. ..+ of unguided TEM plane waves in the last chapter. We will now show in this chapter that many of the characteristics of TEM waves guided by transmission lines are the same as those for asuniform plane wave propagating in an unbounded dielectric medium.

The three most common types of guiding structures that support TEM waves arc:

. a) Parallel-plate transmission line. This.type of transmission line consists of two parallel conducting plates separated by a dielectric slab of a uniform thickness. See Fig. 9-l(a)). At microwave frequencies parillel-plate transmission hncs can be fabricated inexpensively on a dielectric substrate uslng printed-circuit tech- nology. They are often called striplines.

1

' Principles of antennas and radlatmg systems will bd d~scussed In Chapter 11.

9-1 1 INTRODUCTION 371 t

Fig. 9-1 Common types of ~ranslnission iines.

We should note that other wave modes more complicated than the TEhl mode can propagate on 411 three ofthase types ofti:lnsmission lines when the sep;ir;ltiun hetivren the conductors is grci1tr.r ili;in ~crtiiili fri~cliolls of the oper:lLillg IV:IYCICII~III. TI~esc other transmiss.ion modes will be considered in the next chapter.

We will shbw that the TEM wave solut~on ofMaxwell's equations for the parallel- plate guiding ptructure in Fig. 9-!(a) lends directly to a pair of rransmission-line equations. The'generd tra~~rmission-line eq~~;itions C;III also he deriicd ijoill ;I c i rc~~i t model in terms of tile resistance. induclance. conductance, and cap;icit;ince per unit Ierlgth of a line. ~ h d transi~ion from the circuit model to the electromi~gneti~ model is elfected from a network with li~mped-p;iramerer elements (discrete resistors. in- (jllctors. :ind c;~p:~ci to~~s) 10 onc \bi111 d i s~r i i~~~lec l p:ir:i1l~e~crs ( ~ ~ I ~ L ~ I I L I ~ L I ~ c I ih t r i [?~~~io t~~ of R. L, G, a@ k along ih; line). From the transmission-line equations all the chiir- acteristics of &ve propsg;ttion :,long ;, giYen line c;ln be derived ;ind siildied.

The stody 1sJiinlc-li;11-1l?o1iic s~e;itiy-s~;i~c properties of Ira11snlissio11 lillcs is g r ~ ~ i t l y fiicilitaled by thi i~se of grilp1iic:il charls wbich avert the necessity of repe~ited cal- cuhtions with coinpler I I L / ~ : I ~ C ~ S . Thc hest know11 and most ividcly used graphical chart is the Smith churt. The use of Smith cfiart for determining wave characteristics on a transmission line and for impedance-.matching will be discussed.

372 THEORY AND APPLICATIONS OF TRANSMISSION LINES / 9

9-2 TRANSVERSE ELECTROMAGNETIC . WAVE ALONG A PARALLEL-PLATE T RANSMlSSlON LINE

Let us consider a y-polarized TEM wave propagating in the +r direction along a uniform parallel-plate transmission line. Figure 9-2 shows the cross-sectional dimensions of such a line and the chosen coordinate system. For time-harmonic fields the wave equation to be satisfied in the sourceless dielectric region becomes the horno- geneous Helmhol:z's equation, Eq. (8-38). In the present case, the appropriate phasor solution is

E = a,E, = a , . E , e ~ ~ ' . (9-la)

The associated H field is, from Eq. (8-26)

where y and 11 are, respectively, the propagation constant and the intrinsic impedance of the dielectric medium. Fringe fields at the edges of the-plaes are neglected. As- suming perfectly conducting plates and a lossless dielectric, we have, from Chapter 8,

and

. The boundary conditions to be satisfied at the interfaces of the dielectric and the perfectly conducting planes are, from Eqs. (7-52a, b, c. and d), as follows:

At both y = 0 and y = d:

H,, = 0, (9-5)

which are obviously satisfied because Ex = E, = 0 and H, = 0.

Fig. 9-2 Parallel-plate transmission line.

)n along a nal dimen- : tidds the the homo- , t w phasor

n line,

. . , ,, . . 4. ! -!$ . . \<!.;;.;-''y',': :; . ' ,, ' , * : I ' . . . '. '. ,> . , .: > _ . , . ?. ,< . . ' i ' . . . . ; . . , , .

. .. . , . , :,,.

:! \

9-2 1 TEM WAVE ALONG PARALLEL-PLATE LINE 373

At y = 0 (lower plate), a,, = a,:

a, D = psc o r psd = EE, = E E ~ ~ - ~ ~ ~ (9-63)

Eo aj, x H = J,, o r J, , - - a , H , = a , - e-J" . (9 -7a) '7

At y = d (uppet plate), a, = -a,:

E - h y x H = J,, o r J , , , = a z H x = - a z o r - j p i . (9-7b)

r?

Equations (9-6) add (9 -7) indicate thar surface charges and surface currents on rhc conducting planes var) sinusoidally with 2, as d o E, and H,. This is illustrated schematically in Fib. 9--3.

Field phasors E and H in Eqs. (9---12) and (9-1 b) satisfy the two Maxwell's curl equations:

Since E = a,E, and H - a,EI,, Eqs. (9-8) and (9-9) become

and

Ordinary derivatives app~i l r above because phasors E, and H , are functions of .- only.

f

Fig. 9-3 TEM-mode fields, surface charges, and surface currents in parallel-plate transmission line. '

374 THEORY AND APPLICATIONS OF TRANSMISSION LINES 1 9

Integrating Eq. (9-10) over y from 0 to d, we have

'- = ~ L l ( : ) , (9 - 12)

where

v(r) = - J: E, d y = - E, f z )d

is the potential difference or voltage between the upper and lower plates; ' r

is the total current flowing in the +: direction in the upper plate: and

I is the inductance per unit length of the parallel-plate transmission line. The dependence of phasors V ( z ) and I ( z ) on z is noted explicitly in Eq. (9-12) for emphasis.

Siniilarly, wc intcgwtc Eq. (9-1 I ) nvcr .s from 0 to I\ - to obtain

= j o C V ( z ) , where

. is the capacitance per unit length of the parallel-plate transmission line. Equations (9-12) and (9-14) constitute a pair of time-harmonic trai~smissiorz-line

ec1ucrtiot1.s for phnsors V ( z ) and I(:). Thcy may bc coinhincd to yield second-ordcr dilli.rcntial ccluutions ror V ( z ) and for I ( = ) :

The solutions of Eqb. (9-16a) and (9-16b) ate, for waves propagating in the +: direction,

V(Z) = Voe - jsz - (9-17a) and

I(z) = ~ , e - ~ ~ ' , (9-17b)

where the phase conlltant

is the same as that given in Eq. (9-2). The relation between V, and I , can be found by using either Eq. (9 1.1) or Eq. (9- 14) :

which becomes, in view of the results of Eqs. (9-13) and (9-15).

I I 2 ,

which, again, is the same as that of a TEM plahe wave in the dielectric medium

9-2.1 Lossy Parallel-Plate Transmission Lines ----.

Wc l w c so far :i~sulhcd llic parallel-pl~~lc tramnission linc to be lossless. In actual situations loss may arise from two causes. First, the dielectric medium may have a nonvanishing loss tahger.1; and, second, the plates may not be perfectly conducting. To characterize these two effects we define two new parameters: G, the conductance

' This statement will be proved in Section 9-3 (see Eq. 9-87).

376 Ti4EORY AND APPLICATIONS OF TR-ANSMlSSlON LINES 1 9

, . per unit length across the two plates; and R, the resistance per unit length of the two plate conductors.

The conductance between two conductors septlratcd by a dielectric medium having a permittivity E and a conductivity 0 can be determined readily by using Eq. (5-67) when the capacitance between the two conductors is known. We have

Use of Eq. (9-15) directly yields

If the parallel-plate conductors have a very large but finite conductivity a, (which must not be confused with the conductivity a of the dielectric medium), ohmic power will be dissipated in the plates. This necessitates the presence of a nonvanishing axial clcctric field a,E, at the plate sarbces. such that the avera$hynting vector

= aspn =::.Xa(a,E, x a,Hz) (9 -24)

has a y component and equals the average power per unit area dissipated in each of the conducting plates. (Obviously the cross product of a y E , and a ,Hx does not result in a !. component.)

Consider the upper plate where the surface current density is JSu = H,. It is convenient to define a sur/ace impedance of an imperfect conductor. Z,, as the ratio of the tangential component of the electr'c field to the surface current density at the conductor surface.

For the upper plate, we hive E E , =f = A = ~ C 9 (9 -26a)

"SU Hx

where qc is the intrinsic impedance of the plate conductor. Here we assume that both the conductivity 0, of the plate conductor and the operating frequency are sufficiently high that the current flows in a very thin surface layer and can be reprc- xnted by the surlacc current J,,,,, Thc intrinsic impcdvncc of ;l good conductor has been given in Eq. (8-46). We have -

where the subscript c is used to indicate the properties of the conductor.

length oi'the two .

electric~medium lily by using Eq. . We have : '

(9 -23)

~chvity o, (which m), ohmic powcr ):.ivanishing axial ng vector

(9 -23) r-'

ipat - each of ', doe3 not result

s J,, = H,. It is '. Z,, as the ratio nt density at the

\ve assume that 1g frequency are nd can be repre- d c o w t o r has

clor.

I 9-2 1 TEM ~ A V E ALONG PARALLEL-PLATE LINE 377 I

Substihi$ of;Eq. (9-26a) in Eq. (6-24) gives

I (9-27)

, The ohmic poker dissipated in a unit length of :he plate havlng a wldth vj IS s p c , which can be expressed in terms of the total surface current. I = WJ,., as

-. Equation (b-28) is the power dissipated when a sinusoidal current ofdmplitude

I flows through a resistance RJw. Thus, the effective serles resistance per unit length for both plates of a parallel-plate transmission line of wldth iv is

. Table 9-1 lists the expressions for the four distributed purut~~eiers (R, L, G. 2nd C per unit length) of s parallel-plate transmission line of width L I and separation (1.

We note from Eq. (9-26b) !hat surface impedance Z, has a positive renctanc- term X, that is nun3erically equal to R,. If the total complex power (insrend of its real Part, the ohmic Power P.. only) associated with a unit length of the plate is considered. X , will lend to an ititert?aiiniies inductance pcr wit lengtll L, = x,,~,, = A,, ,.,. At hi& frquedcics; Li is negligible in comparison with the exterllal inductlncu ,r.

Table 9-1 Distributed Parameters of Parallel- Plate Transmission Line (Width = W.

Stparation = d )

We note in the calculation of the power loss in the plate conductors of a finite conductivity o, that a nonvanishing electric field a,E, must exist. The very existence of this axial electric field makes the wave along a lossy transmission line strictly not TEM. However, this axial component is ordinarily very small compared to the transverse component E,. An estimate of their relative magnitudes can be made as follows:

I

(9-30)

For copper plates [oc = 5.80 x lo7 (S/m)] in air [s = r0 = 10-'/36n (F/m)] at a frequency of 3 (GHz),

IEJ ~ ~ 5 . 3 x lo-s/Eyl << IE,~. Hence we retain the designation TEM as well as all its consequences. The introduction of a small E: in the calculation of p. and R is considered :I slight pcrturb;ltian.

. -... --- Example 9-1 Striplines consisting of a thin metal strip separated from a conducting ground plane by a dielectric substrate are used extensively in microwave circuitry. Neglecting losses and assuming the substrate to have a thickness 0.4 (mm) and a dielectric constant 2.25, (a) determine the required width w of the metal strip in order for the stripline to have a characteristic resistance of 50 (R), (b) determine L and C of thc line. and (c) detcrminc 11,) along llic linc. (d) Kcpeat parts (;I). (b) ;~nd (c) lor a characteristic resistance of 75 (R).

Solution

a) We use Eq. (9-20) directly to find w.

. - - 0.4 x 377

50 = 2 x (m), or 2 (mm).

a finite xistence ctly not 1 to the made as .

(9 - 30)

ill)] at a

, ~ ~ ~ U ~ t ~ O l l

ti(?n.

n ,r,db circuitry. in) and a ;, In order : I, and C d (c) for a

d

/?

1).

9-3 1 GENERAL TRANSMISSION-LINE EQUATIONS 379 1 . I

d) Since w is inveksely proportional to,&, I we I

have, for Z b = 75 (Q),

$ 9-3 GENERAL TRANSMISSION-LINE EQUATIONS

We will now derive the equations that govern general two-conductor uniform transmission lincs. Trdt~smission lincs clilrcr from ordinary clcctric nctworks in one essential feature. Whercas the physicai dimensions of electric networks are very much smaller than the operating wavelength, transmission lines are usually a considerable fraction ofa wavelength and may even be many wavelengths long. The circuit elements in an ordinary electric network can be considered discrete and as such may be described by lumped parameters. Currents flowing in lumped-circuit elements do not vary spatially over the elements, and no standing Waves exist. A transmission line, on the other hand, is a distributed-parameter network and must be described by

. circuit parameters that are distributed throdghout its length. Except under matched conditions, standing waves exist in a transmission line.

Consider a differential length Az of a transmission line which is described by the following four parameters :

R, resistance per unit length (both conductors), in Qlm.

L, inductance per unit length (both conductors), in Him.

G, conductance per unit length, in S/m.

C, capacitance per unit length, in F/m.

Note that R and ,!, are series elements, and G and C are shunt elements. Figure 9-3 shows the equivalent clcctric circuit olsuch a line segment. The quantit~es o(:, t ) .md v(z + Az, t ) denote the instantkneous voltages at : and z + Az respectively. Similarly, i(z, t ) and i(z + Az, t ) denote the instantaneous currents at z and 2 + Az. Applying ~ i r c h h a s - v o l t a g e law, we obtain

which leads to ,

- &(z, t) v(z + Ai, t ) - 45 t ) = Ri(z, t ) + (9 -30a)

A: at

N i ( z + A r , r ) . .

v(z + Az, 0

Fig. 9-4 Equivalent circuit of a

I-----A~ -----I differential length Az of a two-conductor transmission line.

On the limit as Az -, 0 , Eq. (9-30a) becomes

Similarly, applying KirchhofYs current lilw to the node N in Fig. 9-4, we h:lvc

2o(z + A:, t ) i(z, t ) - G AZV(Z + A:, t ) - C A: ---.

- i(: + A:, t j = 0. (9-32) (71

On dividing by Az and letting Az approach ;era, Eq. (9-32) becomes

I I

Equations (9-31) and (9-33) are a pair of first-order partial differential equations in v(z, t ) and i(z, t). They are the ge~zeral transmission-line equations.'

For harmonic time dependence, the use of phasors simplifies the transmission- line equations to ordinary differential equations. For a cosine reference we write

where V(z ) and I (z) arefinctions of the space coordinate z only, and both may be complex. Substitution of Eqs. (9-34a) and (9-34b) in Eqs. (9-31) and (9-33) yields the following ordinary differential equations for phason V ( z ) and I(z):

' Sometimes referred to as the telegraphist's equations.

9-3.1 Transr

1

nductof

(9-31)

wc l ~ a v e

(9 -33)

n

(9 -33)

yuations

mission- e write

(9-34a)

(9-34b)

i may be 33) yields

(9-35a) m

(9 4

9-3 1 GENERAL TAANSMISSION-LINE EQUATIONS 381

. : Equations (9-3ja) dnd (9-35b) are timq-barionic transmission-line equarions, which - '.reduce to Eqs. (9-19) and (9-14) under lo&led conditions (R = 0, G = 0).

c

I '

9-3.1 ' wave ~haractekistks on an Infinite Transmission Line I

i

The coupled time-harmonic transmission-line equations, Eqs. (9-35a) and (9-35b), can be combined rd solve for V ( z ) and ~ ( z ) . We obtain

I

and

where

is the propagation constccnt whose real and imaginary parts, cc and /,', are the artenrlurm constant (Np/m) and phase constant (radlm) of the line respectively. The nomenclature here is similar to that for plane-wave propagation in conducting media as defined

-,

in Section 8-3. THese quantities are not really constants because, in general, they depend on o in a complicated way.

The solutions df E ~ L . (9-36a) and (9-36b).are

where the plus and minus superscripts dendte waves traveling in the + z and - z directions respectiltely. Wave amplitudes, V; , V; , I:, and I ; are related by Eqs. ,(9-35:l) :~nd (9-35b), mi it is wsy to vcrify (Problcm P.9-5) thot

V,+ V R+joL -- - ---- . lo' I,' y

'

I 1

8 .

382 THEORY AND APPLICATIONS OF TRANSMISSION LJNES 1 9 , .

. ,

~ b r an infinite line (actually a semi-infinitdline with the source at the leA end), the terms containing the eYz factor must vanish. There are no reflected waves; only the waves traveling in the + z direction exist. We have

The ratio of the voltage and the current at any z for an infinitely long line is independent of z and is called the characteristic impedance of the line.

Note that y and Z, are characteristic properties of a transmission line whether or not thc linc is infinitely long. They dcpcnd on R. L, G. C. and to-not on thc Icngth of the line. An inlinitc line simply implics that thcrc arc no rellcq~d waves.

The general expressions for the characteristic impedance in ~<(9-41) and the propagation constant in Eq. (9-37) are relatively complicated. The following three limiting a w s hnvc special significance.

I . Lassless Lirte (R = 0, G = 0).

3) Propagation constant:

c i = O (9-42a)

p = w a (a linear function of o ) . (9 -42b)

b) Phase velocity:

0 1 u =-

p . p=F (constant).

c) Characteristic impedance: l - 7

' - 2. ~ d w - L O S S Line (R << oL, G cc wC). The 1.0~-loss condition is more easily satisfied at very high frequencies.

I ,

IFt end), ; 1 :s; only I

I

is inde-

asily sat-

4

. , ;t, " , ,, . ,,& ;:;g , .., ,:- , . , , . ,, . . . ,.

I . . _ , z , - . . .. ! . i . , , < .!;' ' , 87 "

, v ::, , , , >;: F + : . .". ' ,

'\ 3 ,, ;.

9-3 / GENERAL TRANSM~SSION-LINE EQUATIONS 383

a) Propagation constant: $

+?$+ 2 G.;&)]

P 2 ~ T c (approxinlately a linear function of o)

b) Phase velocit): I 1 u = - - - % - - -

I " /I JLc (approxi~natcly conbtanl).

C) Characteristic impedance:

3. Distortionless L h e (KjL = GIC). If the condition

L. L is satisfied, the expressions for both y and 2, simplify.

a) Propagation constant:

P = om (a linear function of w) .

b) Phase velocity: w 1 u p = - = -

\ B rn (constant). (9 -50)

c) Characteristic impedance:

Ro = $ (constant) (9-51a)

X o = 0. (9-51b)

Thus, except for a nonvanishing attenuation constant, the characteristics of a distortionless line are the same as th6se of a lossless line; na'mely, a constant phase velocity (up = l/m) and a constant real characteristic impedance (2, = R, =j'LF).

A constant phase velocity is a direct conscqtiencc of thc Iinc:~~. tlcpcntlcncc of IIw p l i w co~isl:~nl /I on (I),. Siwc :I si!:11;1I I I S U : I I I J ' c o ~ ~ s i s b 01' ;I I X I I I ~ ol' I ' I ~ ~ ~ I c I I c ~ c ~ , i t is essential Illat thc dill'crcnt l'rcqucncy compoacnts travcl along 'ii-transmission line at the same velocity in order to avoid distortion. This condition is satisfied by a lossless line and is approximated by a line wirh very low losses. For a lossy line, wave amplitudes will be attentuated, and distortion will result when different frequency components attenuate differently, even when they travel with the same velocity. The condition specified in Eq. (9-48) leads to both a constant cr and a constant u p - thus the name distortionless line.

The phase constant of a lossy transmission line is determined by expanding the expression for y in Eq. (9-37). In general, the phase constant is not a linear function of o; thus, it will lead to a up, which depends on frequency. As the different frequency components of a signal propagate along the line with different velocities, the signal suffers dispersion. A general, lossy, transmission line is therefore dispersive, as is a lossy dielectric (see Subsection 8-3.1).

Example 9-2 It is found that the attenuation on a 50-(Q) distortionless transmission line is 0.01 (dB/m). The line has ti capacitance of 0.1 (pF/m).

a) Find the resistance, inductance, and conductance per meter of the line. b) Find the velocity of wave propagation.

c) Determine the percentage to which the amplitude of a voltage traveling wave decreases in 1 (km) and in 5 (km).

Solution

a) For a distortionless line,

(9-50) '

(9-51)

9-51a)

9-51bJ

f ,i dis- I ;:!use --- ., I-, C). 2 of the i\. I t IS

on li,% :d b) e, \4; lx '?

4'JL

cloc1ty. onstant

i ~ n g L I I C unction ent fre- locities, ,I)ersiue,

mission

1g wave n

\

1 . The three relat?ons above are sufficient to solve for the three unknowns R, L, and G in terms bf the given C = 10-lo (F/m):

b) The velocity of wave propagation on a distortionless line is the phase velocity given by Eq. (9-50).

1 1 = 2 x 10' (m/s).

, L J(0.25 x

c) Tho ratio of two voltagcs a distancc z apart along the linc is

After 1 (km), (G/v l ) = e-'Oooa = , -1 ,15 = 0.317, or 31.7%. After 5 (km), (V'/V,) = e-5000a = e-5.75 = 0.0032, or 0.32%.

9-3.2 Transmission-Line Parameters

The electrical properties of a transmission line at a given frequency are completely a

characterized by its four distributed parameters R, L, G, and C. These parameters for a parallel-plate transmission line are listed in Table 9-1. We will now obtain them for ty;w!re and coaxial transmission lines.

Our baslc premile is that the conductivity of the conductors in a transmission linc is asu;illy so !ligli Illill tho dTcct el. tbc wries rcsisli111cc oto lllc ~ ~ l l l p ~ l l i l t i ~ n 01 t11c propilgiltiot~ ct)llsti~ill is ~sgligihlc. Ihc is~pl ic i~ l ion lsinp 1I1:11 ~ l l c wi~vcs on the lilx am approximate'g %M. We may write, ir) dropping R liom Eq. (9-37),

386 THEORY AND APPLICATIONS OF TRANSMISSION LINES / 9 t i i ' ' ,

[ - , . , , From Eq. (8-37) we know that the propagation constant for a TEM wave in a medium with constitutive parameters (p, E, Q) is I ,

But 1 G a , -- - - (9-54) i f C E i

b

in accordance with Eq. (5-67); hence comparison of Eqs. (9-52) and (9-53) yields L c

(9-55)

Equation (9-55) is a very useful relation, because if L is known 'or a,!ine with a given medium, C can be determined, and vice versa. Knowing C, we can find G I

from Eq. (9-54). Series resistance R is determined by introducing a small axial El as a slight perturbation of the TEM wave and by finding theahmic power dissipated

--. -7

in a unit length of the line, as was done in Subsection 9-2.1. Eq~iation (9-55), of course, also holds for a losslcss line. The velocity of. wave

propugation on u 1o.ssle.s.s trun.srni.ssion line, u, = ]/,./LC, therefore, is ecluul to the velocity of propugation, I/&, of unguided plane wove in the dielectric 06 the line. This fact has been pointed out in connection with Eq. (9-21) for parallel-pi: dte lines. I

I

1. Two-wire tr.ansmission line. The capacitance per unit length of a two-wire transmission line, whose wires have a radius a and are separated by a distance D, has been found in Eq. (4-47). We have

I

(9-56)+ 1

cosh- ' ( D l 2 4 D

I I

From Eqs. (9-55) and (9-54), we obtain

L = c o s h ' (5) (Him) 1 I and

L

. i '-7

(9-58)t i

i To determine R, we go back to Eq. (9-27) and express the ohmic power I , dissipated per unit length of both wires in terms of p,. Assuming the current i

+ cosh - ( D l 2 4 z In (Dlu) if ( ~ 1 2 ~ ) ' >> 1.

1 medium.

(9 -53)

(9-54)

I ) yields

(9-55)

line wth , i n Jimi G 11 m a 1 E, t ~ssipated

/-

1 oj \VLlLV

l rol he llrle. This lines.

&Ire trans- nce D, has

(9 -56)'

(9-57)+

n ("8)'

mic power .he current

I- . . ,, . .'Z .; ':F "'""

, ,,. p .- . - r 1 . i :

I . ' , 3 8 i; ,

\ 4 " f ,! ' 9-3 1 GENERAL TRANSMISSION-LINE EQUATIONS 381 ... .

'. i : ' I I 6 L

0 1 )

J , (Aim) to flow. in a very thin surface layer, the current in each wire is I = 2nal,, and . .

(9-59) . I

Hence the seriea resistance per unit length for both wires is

In deriving ~ ~ l ( 9 - 5 9 ) and (9-60), we have assumed the surface current .I, to be uniform over the circumference of both wires. This is an approximation. inasmach . as the proximity of the two wires tends to make the surface current nonun~form.

2. Couxiul rransntlssion fine. The external inductance per unit length of a coaxial transmission line with a center conductor of radius a and an outer conductor of inner radius b has b:xn found in Eq. (6-124):

From Eq. (9-55), we obtain

To determine K, we again return to Eq. (9-27), where JSi on the surface of the center conductor is different from J,. on the inner surface of the outer con- dudor. We mudt haye

388 THEORY AND' APPLICATIONS OF TRANSMISSION- LINES 1 9

Table 9-2 Distributed Parameters of TWO-Wire and . Coaxial Transmission Lines

parameter Two-Wire Line Coaxial Line Unit

I I.1 6 ~ c 0 s h - l (k) -1n-

t

HIm L 2n a n

Internal Inductance IS not included - 1.

From Eqs. (9-65a) and (9-65b), we obtain the resistance per unit length: , , I

(9 -66) I

i The R, L, G, C parameters for two-wire and coaxial transmission lines are listed 1

in Table 9-2.

9-3.3 Attenuation Constant from j

Power Relations i

The attenuation cqnstant dl a traveling wave on a transmission line is the red part I

1 of I ~ C propitgtion constant; it can be determined from the basic definition in Eq. ; , (9-37):

a = &',(I) = :1( [J(R + jmL)(G + jwC)] . . (9-67) ' I The attenuation constant can also be found from a power relationship. The /?

phasor voltage and phasor current distributions on an infinitely long transrnlssion line (no reflections) may be written as (Eqs. (9-4Un) and (9-40b) with thc plu\ snpcr- i

script dropped for simplicity): i I

v ( ~ ) = v o e - ( n + j P ) z . (9-68a) j

I

n llgth,

(9 -66)

~ c s are listed

rhe real part ~ i t ~ o n in Eq.

(9-67)

ons ship. The ~r:~nsmission e ph . W e r - P -

,, -68a)

(9 -68 b)

,: 9-3 / GENERAL TRA~SMISSION-LINE EQUATIONS 389

I . t * r \

The time-average b o a r prbpagated along the bne at any z is ( 6

7 P(z) = +%[V(Z)~*(Z)]

The law of conservation of energy requires that the rate of decrease of P(I) with distance along the lide equals the time-avsrage.power loss PL per unit length. Thus,

= 2ctP(2),

from which we obtain the following formula:

Example 9-3

a) Use Eq. (9-701 to lind thc attenuation constant of a lossy transmission line with distributed paradeters R, L, G and C.

b) Specialize the result in part (a) to obtain the attenuation constants of a low-loss line and of a distortionless line.

Solution

a) For a lossy trahshission line the time-average power loss per unit length 1s

PL(~) = j[lI(zj12R + p ( z ) 1 2 q

- v 2 (R + GIZ~~' )~- ' " . (9-71) -- 2lz0l2

~ubstitutibn of Eqs. (0-69) and (9-71) in Eq. (9-70) gives

(R + GIZOlZ) (NpIm).

b) For a i$;toss line. Z , z R,, = Jq?, Eq. (9-72) becomes

390 THEORY AND APPLlcATloNs OF TRANSMISS~O'N I

, which checks with Eq. (9-45). For a distortionless line, Zo = Ro = m. Eq. (9-72a) applies, and

c x = h & + ; k ) , I

2

which, in view of the condition in Eq. (9-48), reduces to I I

. = R & ,(9-72b)

Equation (9-72b) is the same as Eq. (9-494. ! 9-4 WAVE CHARACTERISTICS ON' FINITE TRANSMISSION LINES

In Subsection %3.l we indk~ ted that the generd solutions for the time-harmonlc one-dimensional Helmholtz equations, Eqs (9-36a) andL9-36b), for transmission

-. lines are (9-73a)

where v: V L & 10' 1,

For infinitely long lines there can be only forward waves traveling in the + .- direction, and the second terms on the right side of Eqs. (9-73a) and (9-73b). representing reflected waves, vanish. This is also true for finite lines terminated in a characteristic impedance; that is, when the lines are matched. From circuit theory we know that a maximum transjer of power fqom a given voltage source to a load occurs under "inuiciied c o ~ ~ d i t i o ~ d ' when the load impedarlce is the complex co~ljugate of the source i~npedn~lce (Problem P.9-1 I). In transmission line terminology, a line is inatci~ed d ~ e n i l ~ r l o d ~ m , w d a ~ ~ r e i s rrlltcrl to thc c~hirrcrc/c~ris/ic irnpedencr (not the compicx corqilqalr [hi. charucteristic impedance) of the line.

I I il i

i---------~. . z = o z = P z'= f z'= 0 I

I Fig. 9-5 Finite transmission line terminated with load impedance ZL. i

. I .

Lct us now consider ihc gcneral c a u of it f h l c transmission line having a characteristic imp&ince-zil terminated in an arbitraty load impedance Z,, as depicted in Fig. 9-5. The lhgth Of the line is C. A sinusoidal voltage source V , E with an internal impedance 2, is conriected to the line at z = 0. In such a case,

which obviously cadhot bc satisfied without the second terms on the right side of Eqs. (9-73a) and (9-33b) unless Z, = Z,. ~ h u s , reflected waves exist on unmatched lines. i.

Given the characteristic y and Zo of the jine and its length &, there are four unknowns V:, V,', f,C, and I, in Eqs. (9-73aj and (9-73b). These four unknowns are not all independent because they are constrained by the relations at : = 0 mcl ,

at z = L. Both V(z) and Iiz) can be expressed either in terms of I/; and I, at the input end (Problem P.9-12), or in terms of the conditions at the load end. Conslder the latter case.

Let z = C in Eqs, (9-Y3a) and (9-7313). We have

VL = V:e-yt + V;eYC (9 -76a)

v,+ J' ; 1, = - e-yc - - Y/ e . z 0

(9-7Gb) 2 0

Solving Eqs. (9-76aj and (9-76b) for V, and V;, we Have

V: = $(VL + I,z,)~Y' - (9-77a)

V f = $(VL Ti ILZo)Ztt, . (9-77b)

Substituting Eq. (9-75) in Eqs. (9-774 and (9-77b), and using the results in Eqs. (9-73a) and (9-73b), we 3btain

Since L and z appear together in the combination (L - z), it is expedient to introduce a new variable z' = G - 2, which is the distance measured backward from the load. Equations (9-78a) aiid (9-78b) then become

We note here that although the same symbols V and I are used in Eqs. (9-79a) and (9-79b) as in Eqs. (9-78a) and (9-78b);the dependence of V(zl) and I(zl) on z' is different from the dependence o f V(z) and I(z) on z.


- <s I I ,

. . - The use of hyperbolic functions simplifies the equations above. Recalling the relations

eY" + e-Y" = 2 cosh yz' and eY" - e-?" = 2 sinh yz',

we may write Eqs. (9-79a) and (9-79b) as

V(zf) = IL(ZL cosh yz' + Z, sinh yz') (9 -80a)

I

(9-80b) . 6

which can be used to find the voltage and current at any point along a transmission line in terms of I,, Z,, y, and Z, .

The ratio V(z1)/I(z') is the impedance when we look toward the load end of the line at a distance z' from the load.

V(z') Z, cosh yz' + Z o sinh yz' Z(zl) = - = zo 4 ~ ' )

1 Z, sinh yz' + Z, cosh yz' (9-81)

' I Z, + Zo tanh yz' Z(zf) = z,

Zo + 2, tanh yz'

At the source end of the line, z' = l, the generator looking into the line sees an input impedance Zi.

Z, + Z, tanh yt' z, = (Z),,, = Zo -- Z ' = J Zo + 2, tanh yL

As far as the conditions at the generator are concerned, the terminated finite transmission line can be rep!aced by Z,, as shown in Fig. 9-6. The input voltage I/;. and input current I, in Fig. 9-5 are found easily from the equivalent circuit in Fig. 9-6.

A- Fig. 9-6 Equivalent circuit for finite transmission line in Figure 9-5 at generator end.

Tll!

teri is s t

the

Ex 7 irc

line loP and

1 .?! ,i .bL ', ,. . . :.. tf: :

, ; . , 1.. b' ' . .. : ' f : ,,;

:callitlg the . : 3 1 @. , : , , I

'., , . : , * ;> : . ,: ., , . ' . . . ,<. &

> ' * . I . .

, : , ' ' ,,'. : . : ''

1 l '

I

1 L.

ransmission

d end of the

sees an input

I finite trans- ~l tage I/, and t id Fig. 9-6.

Of course, the volta& and current at any other location on line cannot be determined by using the equivalent circuit in Fig. 9-6.

The average power delivered by the generator to the input terminals of the line is I i

(PaV)~ = @~[JV?],=O, = I = / . - (9-85)

The average power delivered to the load is

Fur a lossless line, cons~xvation of power requires that (Pa,), = (Pa,),. A particularly impomnt special case is when a line is terminated wlth its charac-

teristic impedance; that is, when Z L = Z,. The input impedance, Z , in Eq. (9-83), is seen to be equal to Z, . As a matter of fact, the impedance of the line loolung toward the load at any distance z' from the load is, from Eq. (9-82),

Z(zf) = Z , (for Z , = Z,). (9-57)

The - oltage and cdirent equations in Eqs. (9-7821) and (9-786) reduce to

V ( z ) = (ILZ,eyL)e-Y" = Xe-?' (9-8Sa) I (z) = ( ~ ~ e ~ ' ) e - ~ ' = I ,e-yz . (9-S8b)

Equations (9-88a) and (3-88b) correspond to the pair of voltage and current equations-Eqs. (9-40a) and (9-40b)-representing waves traveling in +I direction. and them are 110 rcllcctrd WL vcs. Hence, wiletl a jirtite trnnrplnsiurr litle is termitlured wtii its own characteristic itnpedunce (when a jinire transmission line is matched), rhe uoltuge und current dist~ibutions on the line are exactly the same as though the line had been extended to injinity.

Example 9-4 -A signal generator having an internal resistance 1 (Q) and an open- circuit voltage v&t) = 0.: cos 2n108t (V) is connected to a 50 (Q) lossless transmission line. The line is 4 (nl) locg and the velocity of wave propagation on the line is 2.5 x 10' (mh). For a matched load, find (a) the instantaneous expressions for the voltage and current at an arbitrary location on .the line, ( b ) the instantaneous expressions

. . t

, . 394 THEORY AND APPLICATIONS OF TRANSMISSION LINES / 9

for the voltage and current at the load, and (c) the average power transmitted to the load.

Solution

a) In order to find the voltage and current at an arbitrary location on the line, it is first necessary to obtain those at thc input end (z = 0, z' = I). The given quantities are as follows.

5 = 0.3/0" (V), a phasor with a cosine reference Z , = R, = 1 (n) Z0 = Ro = 50 (0) o = 277 x- lo8 (rad/s)

up = 2.5 4 lo8 (m/s)

f = 4 (m).

Since the line is terminated with a matched load. Zi= Z , = 50 (R). The voltage I. and current at the input terminals can be evaluated from the equivalent circuit

in Fig. 9-6. From Eqs. (9-84a) and (9-84b) we have #

50 v=- ' 1 + 5 0

x 0.3/0" = 0 .294p (V)

I.=-- 0'3/0" - O.0059p (A). ' 1 + 5 0

As only forward-traveling waves exist on a matched line, we use Eqs. (9-68a) and (9-68b) for, respectively, the voltage and current at an arbitrary location. For the given line, a = 0 and

Thus, V(z) = 0.294e-j0.8nz (V)

I(z) = 0.0059e-j0.8"z (A).

These are phason. The corresponding instantaneous expressions are, from Eqs. (9 -34a) and (9 -34b),

. I !

tted to ihe . ' ,

* , , I . ,

e line, it is ved quan-

he voltage ent circuit

/-'

1s. (9-68a) location.

from Eqs.

f7

9

Y i b) At the load, z = = 4 (m),

I \ ut4, t) = 0.294 cos (2n!b8t - 3 . 2 ~ ~ ) (V)

:' i(4, t) = 0.0059 cos (2rr108t - 3 . 2 ~ ) (A).

c) The average power transmitted to the load on a lossless line is equal to that at the input t e r h d l s ,

(PJJL = (P.Ji = :!%'L[v(z)1*(2)]

= j(0.294 x 0.0059) = 8.7 x (W) = 0.87 (mW).

9-4.1 Transmission Lines as Circuit Elements

' I ,

Not only can transdission lines be used zs wave-guiding structures for transkrring power and informatibn fium onc point to another, but at ultrahigh frequencies- UHF: from 300 ( M ~ z ) t.) 3 (GHz); wavelength, from I (m) to 0.1 (m)-they may scrvc 21s circuit clcmanls. A t thcsc rrcqucncics, ordinary lumped-circuit clcmcnts arc dilficult to make, and str.ly fields become important. Sections of transn~ission lines can be designed to gibe an ,inductive or capacitive impedance and are used :o match

'

an arbitrary load td the internal impedance of a generator for maximum power transfer. The required length of such lines as circuit elements becomes practical in the UHF range.

In most cases transmission-line segmenti can be considered lossless. y = jjl. Z , = R, and tanh 7L' = tanh (j/V) = j tan /If. he formula in Eq. (9-83) for the input impedance Z i of a lossles line of length G terminated in 2, becomes

ZL + jRo tad /?/ Zl = R, ------------

Ro + jZL tan /)/

(Lossless line)

We now consider several important special cases.

1. Open-circuit rerntinatm (2, -t a). We have, from Eq. (9-89),

396 THEORY AND APPLICATIONS OF TRANSMISSLON LINES I 9 /

Fig. 9-7 Input reactance of open-circuited transmission line.

P

Table 9-3 Input Reactance of Open-Circuited Line, X,,

1 1 K ( 2 n - 1 ) - < L < n -

4 2 (2n - 1) - < Pd < nn

2 Inductive

I. IE (2n - 1) -

4 (2n - 1) -

2 0

When the length of an open-circuited line is very short in comparison with a wavelength, p.4 << 1, we can obtain a very simple formula for its capactive reactance by noting that tan P.4 2 pf. From Eq. (9-90) we have

which is the impedance of a capacitance of C.4 farads.

2. Short-circuit termination (2, = 0). In this case, Eq. (9-89) reduces to

Zi = jX,, = jRo tan b.4. (9-92)

Since tan p.4 can range from - co to + co, the input impedance of a short-circuited lossless line can also be either purely inductive or purely capacitive, depending on the value of p.4, Figure 9-8 is a graph of Xis versus 4, and some important properties of Xis are listed in Table 9-4. ,

parison with pactive reac-

ort-L i t e d :, depending le important

Fig. 9-8 Idput reactance of short-circuited transmission line.

A /. ( 2 n - 1 ) - < r ; < n -

4 2 Capacitive

It is instructive to note that in the range where X,. is capacitive X,, is inductive, and vice versa. The input reactances of open-circuited and short-circuited lossless transmission lines are the same if their lengths differ by an odd multiple of 214.

When the length of a short-circuited line is very short in comparison with a wavelength, DL << 1 Eq. (9-92) becomes approximately

which is the irnpcd~ncc of +n inductilncc of LL henries. 3. Quarter-wave secti~ns (f = 114, DL = ~12). When the length of a line is an odd

mult$l+pf A/4, l = ( 2 , - 1)1/4, ( n = 1,2,3, . . .),

Hence, a quarter-wave lossless line transforms the load impedance to the input terminals as its inverse multiplied by the square of the characteristic resistance; it is often referred to as a quarter-wave transformer. An open-circuited, quarter-wave line appears as a short circuit at the input terminals, and a short-circuited quarter- wave line appears as an open circuit. Actually, if the series resistance of the line itself is not neglected, the input impedance of a short-circuited, quarter-wave line is an impedance of a very high value similar to that of a parallel resonant circuit.

4. Half-wuve sections (C = 142, P t = n). When the length of a line is an integral multiple of i.12, d = n1/2 (11 = 1 , 2, 3, . . .),

Equation (9-95) states that a half-wave lossless line transfers the load impedance 1 to the input terminals without change. I

i

t anp t = O , 1 and Eq. (9-89) reduces to t

1

Open- and short-circuit terminations are easily provided on a transmission line. j By measuring the input impedance of a line section under open- and short-circuit I

conditions, we can determine the characteristic impedance and the propagation , constant of the line. The following expressions follow directly from Eq. (9-83). I

Zi = ZL (Half-wave line).

Open-circuitrd line, 2, -, co :

(9-95) :

Zi,, = Zo coth yd.

Short-circuited line, ZL = 0:

Z, = Zo tanh ye..

From Eqs. (9-96) and (9-97) we have

(9 -94)

J , r / ~ e input sistunce; it drter-wave 2d quarter- ol ihe line r - ~ a v e line aut circuit.

dn integral

P.'

(9-95)

1 irnpcdarm

nlsslon line. ,hort-circuit xopagation (9-83).

(9 -96)

h 9 7 )

(9 -98)

and I

, (9 -99)

Equations (9-98) ahd (9.199) apply whether Or not the line is lossy.

Example 9-5 The open-circuit and short-circuit impedances measured at the input terminals of a very low-loss transmission line of length 1.5 (m), which is less than a quarter wave~en~td, itre respectively -fi4.6 $2) andj103 (R). (a) Find Z, and y of the line. (b) Without ckanging the operating l'requency, lind ihe input impedance of a short-circuited lind that is twice the given length. (c) How long should the shon- circuited line be id ordeifor it to appear as an open circuit at the input terminals?

Solution: The givbn quantities are

Zit, = -j54.6, % , = jlO3, P = 1.5 ,

a) Using Eqs. (9-98) and (9-99), we find

Z , = ,/ -j54.6(]103) = 75 (Q)

1 7 = - tanh-l /= = - 1,s

j tan- ' 1.373 = jO.628 (rad/m). -j54.6 1.5

b) For a short-circuited line twice as long, I = 3.0 (m),

ye = j0.628 x 3.0 = j1.884 (rad).

The input impedance is, from Eq. (9-97),

Z,, = 75 tanh (j1.884) = j75 tan 108"

= j75( - 3.08) = -j231 (R).

Note that Zis for the 3 (m) line is now A capactive reactance, whereas that for the 1.5 (m) line in p x t (a) is an inductive reactance. We may conclude from Table 9-4 that 1.5 (m) < 4 4 < 3.0 (rn).

c) In order for a short-circuited line to appear as an open circuit at the input terminals, it should be an odd multiple of a quarter-wavelength long.

. 2z 2n A = - - = - l o (m). p 0.628-

Hence the required h e h g t h is

400, THEORY AND APPLICATIONS OF TRANS'MISSION ~ ~ N E S 1 9 , . ' , , ,

. , i : I

9-4.2 Lines with Resistive Termination a ! : ,

When a transmission line is terminated in a load impedance ZL different from the characteristic impedance Z,, both an incident wave (from the generator) and a re- : . ! flected wave (from the load) exist. Equation (9-79a) gives the phasor expression for the voltage at any distance z' = I - z from the load end. Note that, in Eq. (9-79a). the term with eYz' represents the jncident voltage wave and the term with e-"' represents the reflected voltage wave. We may write

,

where

is the ratio of the complex amplitudes o f the reflected and incident voltage waves at I 1 i the load (z' = 0) and is called the voltage rejection coeficient of the load impedance

Z,. It is of the same form as the delinition of the reflection coeficient in Eq. (8-93) 1 for a plane wave incident normally on a plane interface between two dielectric media. It is, in general, a complex quantity with a magnitude Irl 5 1. The current equation !

corresponding to V(zl) in Eq. (9-100a) is, from Eq. (9-79b). i

The current reflection coefficient defined as the ratio of the complex amplitudes of the reflected and incident current waves, I ; / I , + , is different from the voltage reflection coefficient. As a niatter of fact, the former is the negative of the latter, inasmuch as l ; / l : = - V,/V:, as is evident from Eq. (9-74). In what follows, we shall refer only to the voltage reflection coelkicnt

For a lossless transmission line, y = jfl, Eqs. (9-100a) and (9-100b) become

I V ( z l ) = 4 2 ( Z , , + R,,)ejp"[l + re- jZ6"]

rent from the tor) anii a re- :xpression for , 1 Eq. (9-79a), with e-Yz' re-

(9-100b)

:s amplitudes le voltage re- ter, inasmuch we shall refer

I become

b The voltage a n current phasors on a lossless line are more easily visualized : from Eqs. (9-80a) ahd (9-80b) by setting y =ID and VL = I,ZL. Noting that cosh j0 =

cos 8, and sinh j8 31 j sin 8, we obtain -. .

,V(zl) = VL cos pz' 4- jILR, sin pz' (9-1033)

I(,-') = IL cos pi' + j - sin /3z1. (9-103b) L

(Lossless line) If the terminating impedance is purely resistive, 2, = RL, VL = ILRL, the voltage and current magdituderi are given by

, 1 V(z')J = vL Jcos2 O:' + (R,/R,)' s i n 2 F (9 - 1 04a) (I(:')l = I ~ J C O S ~ /I:' + (RL/R0)2 s i n 2 p ,

- -. (9-104b)

where R, = ~ L / C . Plots of IV(zt)l and )l(zr)I as functions of ;' are standing waves with their maxima Bnd minima occurring at fixed locations along rhc ilnc.

Analogously to t h ~ pl;il~c-wi~ve caw in Eq. (8-100). we dcfinc the r:ltio of thc rn:lrlmum to the min in~u~n voltages along a linite, ternlinated lille iis the rialding-mce ratio, S:

(D~mcnsronlcss). d 1 - 11-1 (9- 105)

The inverse relatiod of eq. (9-105) is

(9-106)

I It is clear from Eqs. (9-105) and (9-106) that on a lossless tr;lnsmission line

r = 0, S = 1 when ZL = Z , (Matched load); I-= -1, S-60 when ZL = 0 (Short circuit); r = + 1, S -+ when ZL - m (Open circuit).

Because of the wide range of S, it is customary to express it on a logarithmic scale: 20 log,, h!!dB). Sfbnding-wave ratio 5 defined in terms of lIma,l/~~,,,lnl results in the same expression as that defined in terms of 1 V,,,//(V,~~,( in Eq. (9-105). A high standing- save ratio on iI line is i ~ ~ i d c s i ~ ~ h l c lxa i t~sc i t results in a 1;lrgc power loss.

Examination of Eqs. (9-lO2a) and (9-lO2b) reveals that /VmaX/ and II..,,/ occur &ether when ( 1 kz'( = 1, independent O~Z.) :

402 THEORY AND APPLICATIONS OF T RANSMlSSlON LINES 1 9

On the other hand, Iv ; , ,~ and (I,,,I occur together when'

8,-2pzk= -(2n+ 1)n, ( n = O , 1,2,.:.). (9-108) ; .-.

For resistive terminations on a lossless line, 2, = RL, 2, = R,, and Eq. (9-101) sim~lifies to

RL - Ro (Resistive load). r = RL + Ro

The voltage reflection coefficient is therefore purely real. Two cases are possible.

1. RL > R,. In this case, r is positive rcal and 0, = 0. At the termination. 2' = 0, and condition (9-107) is satisfied (for II = 0). This mc:lns thi11 ti vo1t;igo ~n;lxi~num (current minimum) will occur at the terminating resistance. Other rnaxil~xi of tllc voltagc stunding wavc (minitnu of the currcnt st;~nding w;ivc) will hc locatcd i i l 2/3z1 = 2rm, or z' = ni42 (11 = 1,2, . . .) from the load.

2. RL < R,. Equation (9-109) shows that r will be negative red and 0,. = -n. At the termination, :' = 0, und co~~di t io~l (0 10s) is sntislicd (for 11 = 0). A voltogc minimum (current maximum) will occur at the terminakng resistance. Other minima of the voltage standing wave (maxima of the current'standing wave) will be located at z' = rd/2 (11 = 1,2, . . .) from the load. The roles of the voltage and current standing waves are interchanged from those for the case of RL > R,.

Figure 9-9 illustrates some typical standing waves for a lossless line with resistive terminaticn.

. The standing waves on an open-circuited line are similar to those on a resistance- terminated line with RL > R,, except that the IV(zl)/ and Il(zl)l curves are now magnitudes of sinusoidal functions of the distance zr from the load. This is seen from Eqs. (9-104a) and (9-104b), by letting RL + co. Of course, I, = 0, but VL is finite. Wc have

I v(zt)( = VL lcos Vz ' ( (9 - 1 lua)

All the minima go to zero. For an open-circuited line, r = 1 and S -+ co.

Fig. 9-9 Voltage and current standing waves on resistance-terminated lossless lines.

ossible.

ion, z' = 0, : maximum. xima of the : located at

r = -rr.At i. X voltage ,lace. Other

wave) will i o l t a q n d

i It,.

d rcsistance- re now mag- is seen from t V, is finite.

I V(z')I for open-c~rcu~ted line. - IMz1)l for short-circuited h e ,

L---- ll(z')l for open-c~rcuitcd line. ' 1 V(-zl)/ for short-circuiml h e .

Fig. 9-10 Voltage h d current standing waver on open- and shortsircu~ted lossless lines. ,i .

On the other and, the standing waves on a short-circuited line are simdar to i; those on a resistance-terminated line with RL < R,. Here RL = 0. VL = 0, but 1 , is finite. Equations (9-104a) and (9-104b) reddce to

I V(zf)l = ILRo /sin Pz'l (9-1 1 la) Il(z')/ = I , /cm bz'l. (9 - l l lb )

Typical standing Wzves for open- and short-circuited lossless lines are shown in Fig. 9- 10.

Example 9-6 The staxling-wave ratio S on a transmiss~on line is an easily measurable quantity. (a) Show how the value of a terminating resistance on a lossless line of known charactetistic impedance Ro can b determined by measuring S. (b) What is the impedance of the line looking toward the load at a distance equal to one quarter of the operating wavelength?

Solutio)~ :

a) Since the terminating impedance is purely resistive, ZL = R,, we can determine whether RL is greater than R, (if there are voltage maxima at z' = 0, i/2, i, etc.) or whether RL is less than R, (if there are voltage minima at ;' = 0, i.12, 1, etc.). This can be easily ascertained by measurements.

First, if RL > R,, 0, = 0. Both /VmaXI and (1.~~1 occur at pz' = 0 ; and 1~ ,,,,,I and (1,,.1 occuk at /Irt = n/2. We have, from Eqs. (9-102a) and (9-1OZb).

404 THEORY AND APPLICATIONS OF TRANSMISSION

Second, if RL < R,, 8, = -A . Both (Vmi,I and /1,,;I occur at pz' = 0; and /vmaxl and IImi,I occur at pz' = n/2. We have

b) The operating wavelength, A, can be determined from twice the distance between two neighboring voltage (or current) maxima or minima. At 2' = 214, 11:' = nI2. cos /kt = 0, and sin /I:' = 1. Equations (9-103a) and (9-1UbJ become

V(jL/4) = jILRo

(Question: What is the significance of thc j in thcsc cquations?) The ratio of V(1.14) to 1(1/4) is the input impedance of a quarter-wavelength, resistively terminated, lossless line.

This result is anticipated because of the impedance-transformation property of a quarter-wave line given in Eq. (9-94).

9-4.3 Lines with Arbitrary Termination

. In the preceding subsection we not:d4hat the standing wave on a resistively terminated lossless transmission line is such that a voltage maximum (a current minimum) occurs at the terminatiorwhere z' = 0 if RL > R,, and a voltage minimum (a current maximum) occurs there if RL < R,. What will happen if the terminating impedance is not a pure resistance? It is intuitively correct to expect that a voltage maximum or minimum will not occur at the termination, and that both will be shifted away from the termination. In this subsection we will.show that information on the direction and amount of this shift can be used to determine the terminating impedance.

le ratio of brivel~f ler-

operty of a

rzmtirrared ti[,) OF

'rent 1-

.iIlCC is 1 m t

x i or away from : direction mce.

Let the termindting (or load) impedand be 2, = RL + jXL, and assume the voltage standing wdve on the line to look like that depicted in Fig. 9-11. We note that neither a voltage maximum noca voltage minimum appears at the load at z' = 0. If we let the standing wave continue, say, by an extra distance f,,,, it will reach a minimum. Thc volthge minimum is where it should he i f the original termmating impedance ZL is replaced by a line section of lehgth l,,, terminated by a pure resistance . R, < R,, as shown in the figure. The voltage distribution on the line to the left of the actual termination (where z' > 0) is not changed by this replacement.

The fact that any complex impedance can be obtained as the input impedance of a section of loul&s line terminated in a resfstive load can be seen from Eq. (9-89). Using R, for 2, add dm for d, we have '

I

The rcll u d im:&ar) pails of Eq (9-114) form two equations, Sroin w h ~ h t l x two unknowns, R,, hnd I,,, can bc solved (sce Problem P.9-34).

Thc load impcd:ln~c % , can bc determined expcrimcntally by musunng the stand~ng-wave ratio S and the distance 2: in Fig. 9-11. (Remember that 1:. + i, = ;-,Q.) The procedure is as follows:

S - l 1. Find lrl from S. Usu Il-1 = -- from Eq. (9-106).

S + l 2. Find 0, from 6 Use 0, = 2P::, - n forjl = 0 from Eq. (9-10s).

t

I ' Fig. 9-11 Voltage standing wave on line ------I

t;n z 'Z o I a terminated by arbitrary im~edance and - . 1.- equivalent line section with pure resistive -7 load.

A Find ZL, which is the ratio of Eqs. (9-102a) and (9 - 102b) at I' = 0:

1 + ITlejer ZL = RL + FL = Ro

I - I r l ,or - (9-115)

The value ;f Rm that, if terminated on a line of length em, will yield an input impedance ZL can be found easily from Eq. (9-1 14). Since R, < Ro. R,,! = R&.

The procedure leading to Eq. (9-1 15) is used to determine Z, from a measurement of S and of 2;. the distance from the termination to the first voltage minimum. Of course, the distance from the termination to a voltage maximum could be used instead of 2:. However, the voltage minima of a standing wave are sharper than the voltage maxima. The former, therefore, can be located more accurately than the latter, and it is preferable to find unknown quantities in terms of S and &.

Example 9-7 The standing-wale ratio on a lossless 50-(Q) trvnsmission line terminated in an unknown load impedance is found to be 3.0. Thc dintan& betwccn successive voltage minima is 20 (cm), and the first minimum is loc;~tcd i l l 5 ( ~ 1 1 1 )

fro111 the load. D ~ ~ ~ I I I I I I ~ (:\I ~ I I C ~cllcctio~\ cocIlicic111 l', : I I I ~ (b) [ I I C l ~ d U I I ~ C L I J I I C C Z,. In addition, find (c) the equivalent.length and termina&@-resistance of a line such that the input impedance is equal to 2,.

Solution

a) The distance between successive voltage minima is half a wavelength.

2x 2x - A = 2 x 0.2 = 0.4 (m), /? = - = - = 5 x (rad/m) . i, 0.4

Step 1 : We find the magnitude of the reflection coefficient, lrl, from the standing- wave ratio S = 3.

Step 2: Find the angle of the reflection coefficient, B,, from

0, = 2& - n = 2 x 5n x 0.05 - n = - 0 . 5 ~ (rad)

r = (rider = 0.5e-j0.5n = -j0.5

b) The load impedance ZL is determined from Eq. (9-115):

/ c) Now we find R, and C,,, in Fig. 9-1 1. We may use Eq. (9-1 14)

30 - j4O = 50 Rm + j50 tan pt, 50 + jR, tan pt,

(9 I S ] 5 )

an input RoIS. lsurement irnum. of 1 be used r than the the lattei-,

n line ter- c bctwccn

< i t 5 (crnr mpedancc : of 4 4 e

: standing-

/7

1

9-4 1 WAVE CHARACTERISIICS Ohl FINITE TRANSMISSION LINES 40 i I

and solve the airndtaneous equations obtained from the real and imaginary pans for R; and Pt,,,, Actually, we know i, + t,,, = 112 and Rm = R,/S. Hence,'

1 e m = - - 2

zk = 0.2 - 0.05 = 0.15 (m)

and 50

Rm=- = 16.7 (R). 3

9-4.4 Transmission-Line Circuits

Our discussions on the properties of trmsrnission lines so far have been restricted primarily to the effects of the load on the input impedance and on the characteristics of voltage and current waves. No attention has been paid to the generator at the "other cnd," which is thc sourcc of Lhc wavcs. Just as the constraint (the boundary condition), V, = I&,, which the voltage V, and the current I , must satisfy at the load end (z = l, z' i 0), a constraint exists at the generator end where 2 = 0 and 2' = i. Let a voltage generator V, with an internal impedance 2, represent the source connected to a finite transmission line of length L that is terminated in a load impcdance Z,, as shown i l l Fig. 9-5. Thc additional constraint at : = 0 will cnable the voltage and current anywhere on the line to be expressed in terms of the source characteristics (V,, Z,), the line characteristics (1, Z,, 4, and the load impedance iZ,).

The constraint Bt z = 0 is = V, - I i Z , .

But, from Eqs. (9-i00aj and (9-IOOb),

IL = y (ZL + Z,)eiLII + re-'?' 1 (9-117a)

and

Substitution of Eqg. (9- 117a) and (9-1 17b) in Eq. (9-116) enables us to find

' Another set of solutiofls to pirt (c) is fm = t',,, - 114 = 0.05 (m) and Rm = SR, = 150 (a). Do you see why?

408 THEORY AND APPLICATIONS OF TRANSMISSION

+' is the voltage reflection coefficient of the generator end. Using Eq. (9-118) in Eqs. (9-100a) and (9-100b), we obtain

Similarly,

f

Equations (9-120a) and (9-120b) are analytical phasor expressions for the voltage and current at any point on a finite line fed by a sinusoidal voltage s'ource V,. These are rather complicated expressions, but their significance can be interpreted in the following way. Let us concentrate our attention on the voltage equation (9-120a); obviously the interpretation of the current equa&&-(9-120b) is quite similar. We expand Eq. (9-120a) as follows:

where

V; = r(VMe-yd)e-yz' (9-121b)

V z = rg(TVMe-2Yd)e-Yz. (9-121c)

- The quantity

v - v,zo -z0 + Zg , (9-122)

is the complex amplitude of the voltage wave initially sent down the transmission line from the generator. It is obtained directly from the simple circuit shown in Fig. 9-12(a). The phasor V: in Eq. (9-121a) represents the initial wave traveling in the

Lrr thc volt- 2 source Vg. interpreted

&e equation 3b) is quite -

sansr~~~ssion iown in Fig.

Fig. 9-12 A transffiission4ne circuit and travelitlg waves.

+ z direction. Before this wave reaches the load impedance Z,. it sees Z, of the line as if the line were h n i t e l y long.

When the first wave VT = Vb,e-'' reaches 2, at 2 = d', i t is reflected because of mismatch, resulting in a wave V ; with a complex amplitude T(V,,,e-:IC) travelifig in the - 2 direction. As the wave V ; returns to the generator at z = 0, it is again reflected for 2, # Z,, giving rise to a second wave V: with a complex amplitude T,,ITV,,:,,C-~"') traVclin~ i n 4: direction. This proccss continues indefinitely with refections at both ends, and the resulting standing wave I/(:') is the sum of all the waves traveling in both dircctions. This is illustrated schematically in Fig. 9-lZ(b). In practice, 7 = ct + j p has a real part, and the attenuation effect of r - " l diminishes the amplitude of rl reflected wave each time the wave transverses the length af the line.

When the lirle is terminated with a nlatched load, 2, = Z,, r = 0, only V ; exists, and it stops at the matched load with no reflections. If 2, # 2, but Z, = 2, (if the internal idpedance of the generator is matched to the line), then r f 0 and r, = 0. As a consequence, both V : and V ; exist, and V ; , V ; and all higher-order reflections vanish,

Esnmple 9-8 A 100-(Mllr,) gcnerutor with V, = 100- (V) and ~nternal resistance 50 (R) is connected to J lassless 50 (R) air line that is 3.6 (rn) long and term~narsd in a 25 + j25 (R) load. Find (a) V(z) at a locatiod z Irom the generator, (b) I.; at the input terminals and V, idt tht load, tc) the vo!tage standing-wave ratio on the line, and (d) -. ' the average power delivered to the load. -1-

Ro = 50 (Q), ZL = 25 + j25 = 35.36/450 (R). t' = 3.6 (m).


Thus, w 2n1O8 2n p=-=-- - - (rad/m), p/ = 2.4n (rad) c 3 x lo8 3 - -

r = ZL - Zo - - (25 + j25) - 50 - -25 + j25 - 35.36/135" - Z, + Z, (25 +'j25) + 50 - 100 + j25 103.1/14"

= 0.343/121" = 0.34310.672~

1-, = 0. . a) From Eq. (9-120a), we have

V(z) = - z, + z,; - -- e- j2n:/3[l + 0.343~j(0.672-&8)n j4nzI3

100 e I

We see that, because r, = 0, V(z) is the superposition of only two traveling waves, V : and V ; , as defined in Eq. (9-121).:

b) At the input terminals,

Vi = V(0) = 5(1 + 0.343e-J0.'28n ) = 5(1.316 - jO.134) = 6.611- 5.82" (V) .

At the load, VL = V(3.6) = 5[e-j0.4n + 0.343ej0.272*]

= 5(0.534 - j0.692) = 3.461- 52.3" (V).

c) The voltage standing-wave ratio is

d) The average power delivered to the load is

It is interesting to compare this result with the case of a matchcd load when ZL = ZO = 50 + j O (a). In that case, l- = 0,

V IV,l = Iq ='--I = 5 (V),

2

9-5 / THE SMITH CHART 41 1 /

and a maximum avirage:bower is delivered td the load:

I v: 5 t Mlximum P,, = - = -- = 0.25 (W), 'I 2R, 2 x 50

which is consideradly larger than the Pa. calculated for the unmatched load in part (d).

ThE SMITH CHART , b

Transmission-line ~alculations-such as the determination of input impedance by Eq. (9-89), reflection coefficient by Eq. (9-101), and load impedance by Eq. (9-1 15)-often invdhe tedious miinipulations of complex numbcrr This lcdl~rm c.in be alleviated by using a grilphicd mec!i~d of solut~on. The beit known m d most widely used graphicel chart is the Siriirh cbnrt devlsed by P. H. Smlrh.' Stated suc- cinctly, a Sn~itll chilrt I: a graphical plot of norn~alized resistance and reactance functions in the reflection-coefficient plane.

In order to understlnd how the Smith chart for a l o ~ ~ l c s ~ transmlision h e 15

constructed, let us examine the voltage reflection coefficient of the load impedance defined in Eq. (9-101):

Let the load impedance ZL be normalized with respect to the characteristic impedance R, = of the line.

where r and x are the normalized resistance and normalized reactance respectively. Equation (9-101) cdn be rewritten as

where r, and r, are the real and imaginary parts of the voltage reflection coefficient r respectively. The inverse relation of Eq. (9-124) is

-

P. H. Smith. "Transmission-linc calculator," Electronlo. vol. 12, p 29. January 1939: m d "An improved transmission-line calculator," Electronics, vol. 17, p. 130, January 1944.

412 THEORY .AND APPLICATIONS OF TRANSMISSION LINES I 9

Multiplying both the numerator and the denominator of Eq. (9-126) by the complex conjugate of the denominator, and separating the real and imaginary parts, we obtain

If Eq. (9-127a) is plotted in the?, - Ti plane for a given value of r, the resulting graph is the locus for this r. The locus can be recognized when the equation is rearranged as

. Fig. 9-13 Smith chart with rectangular coordinates.

, . 8 . : . , i ;

; , , . . , ! .

: complex . . ' . / .

, , . ; I : weobtain . 1 ; , : ! :

/ .

(Y-I&) . , , , ,

. . : I

(9-l27b)

. ' 5

: redulting , ;, 1 ., .

tion is re- , . : 1

. . I .. , fi 9-5 1 THE SMITH CHART 413 1

Several sa~iknt properties of the r-dircleslare noted as follows:

1. The centers oh11 rhrcles lie on the'ir-$xis.

2. Thc r = ~,.circ~c, hwiog ;I unity radi'us and ccntcrcd at thc origin, is thc largcst. - 4 '

I 3. The r-cmles becoAe progressively 'smiller as r increases from 0 toward m, ending at the (r, = 1, Ti = 0) point.

4. All r-circles pdss through the (I-, = 1, Ti = 0) point.

Similarly, Eq. (9- l2lb) may be rearranged as

This is the equation for a circle having radius l / [x( and centered at T, = 1 and T, = l/x. Different values o fx yield circles of different radii with centers at different positions on thc r, = 1 line. A family of the portions of x-circles lying inside thc /I-/ = 1 boundary are shown in dashcd lines in Fig. 9-13. The following is a list of several salient properties of the u-circles.

1. The centers of ali s-circles lie on the Tr = 1 line; those for x > 0 (inductive reactance) lie above the r,-axis, and those for x < 0 (capacitive reactance) lie below the r,-hxis.

2. The x = 0 circle bzcomes the rr-axis.

3. The x-circles become progressively smaller as 1x1 increases from 0 toward a, ending at the [ r , = 1, Ti = 0) point.

4. All x-circles pass through the (T, = 1, rl = 0) point.

A Smith chart is a chart of r- and x-circles in the T, - I-i plane for Irl 5 1. It can be proved thi; the r- and s-circles are everywhere orthogonal to one another. The intersection of an r-circle and an x-circle defines a point that represents a normalized load impedance z , = 1. + ,js. Thc actual load impedance is Z , = R,(r + js). Since a Smith chart plots the normalized impedance, it can be used for cn1cul;~tions concerning :I Inssicss i~x~lklllissitl~l linc wit11 ; I I ~ ;~~.l>iilx~'y ~ I ; I I . : I C ~ C S ~ X ~ ~ L ' ~ ! ~ ~ ~ C L I : I I ~ C C .

As an illustration, point P in Fig. 9 -13 is the intersection of the r = 1.7 circle and the x = 0.6 circle. Herice it represents 2,. = 1.7 + j0.6. The point P,, at (T, = - 1, T i = 0) corresponds toi r = 0 and x = 0 nhd, therefore, represents a short-circuit. The point Po, at (T, = 1, Ti A 0) corresponds to an infinite impedance and represents an open-circuit.

The'Smith chart ic Fig. 9-13 is marked with T, and Ti rectangular coordinates. The same chart can be marked withpolar coordinates, such that every point in the r-plane is specified by a magnitude I f 1 and a phase angle 8,. This is illustrated in Fig. 9-14, where several ]TI-circles are shown in dotted lines and some Or-angles are marked aroufid the Irl = 1 circler The Irl-circles are normally not shown on commercially available Smith charts; but once the point representing a certain

. ,. ' I , I :

I:.'_ -

41 4 THEORY AND APPLICATIONS OF TRANSMISSION

9n0

270'

Fig. 9-14 Smith chart with polar coordinates.

z, = r + jx is located, it is a simple matter to draw a circle centered at the origin , through the point. The fractional distance from the center to the point (compared

i I I

with the unity radius to the edge of the chart) is equal to the magnitude Irl of the load reflection coefficient; and the angle that the line to the point makes with the real axis is 0,. This graphical determination circumvents the need for computing by Eq. (9-124).

Each Irl-circle intersects thc real axis at two points. In Fig. 9--14 we designate the point on the positive-rcal axis (OfJ,,,) as I', and thc point on thc: ncgativc-rcal axis (OP,,) as P,. Since x = 0 along the real axis, l', and P, both represent situations with a purely resistive load, Z , = R,.. Obviously RL > R, at P,,, where r > 1 : and

I RL < I<, a t l',,, where r < 1. ln Eq. (9-1 19) we found that S = K,/R, = r Tor K L > R,. This relation enables us to say immediately, without using Eq. (9-105), that the value of the r-circle passing through the point P , is numerically eqlial to the standing-wave ratio. Similarly, we conclude from Eq. (9-113) that the value of the r-circle

I i A

passing through the point P, on the negative-real axisis numerically equal to 11s. For thc :, = 1.7 + jO.6 point, markcd P in Fig. 9 14, wc find Irl = atid 0,. = 28 . At I,

P,, r = S = 2.0. These results can be verified analytically. '

I In summary, we note the following:

1. All Il-1-circles are centered at the origin, and their radii vary uniformly from 0 to 1. I I

.i i.,. . . 9-5 1 THE SMITH CHART 1415

/-

he origin :.~mpdred ,I'l of the

, ~11th the iputing r

designate a u k e-real i~tuatlOnS > 1 : and R , > R,. . that the he uartd- 'le r-circle I 1,s. For = 1 8 ' 7

3mOto 1.

, 2. The angle, meisured from the positive r h l axis, of the line drawn from the origin through the pdint representing z, equah 0,.

3. The value of the r-circle passing thrdugh the intersection of the irl-circle and tHe positive-real akis eqhals the standing-whve ratio S. I

So far we have based the construction of the Smith chart on the definition of the voltage reflection c o e a e n t of the load irnpdtiance, as given in Eq. (9-101). The input impedance looking toward the load at a distance z' from the load is the ratio of V ( t ' ) and I(z'9. From Eqs. (9-100a) and (9-100b) we have, by writing jp for y for a lossless - *

! , * I I

line. ,

The normalized inbut impedance is

We note that Eq. (9-130) relating z, and = / r l eJ9 is of exactly the same form as Eq. (9-125) rcliitinf ;, and = IT[eJnr. In fact, the laltcr 1s a s p c c ~ l case ol. the former for z' = 0(4 = Or). The magnitude, ~rl, of the reflection coefficient and, therefore, the standing-have ratio S, are not changed by the additional line length ?. Thus, just as we can use the Smith chart to Rnd IrI and Or for a given ;,at the load, we can keep Irl constant and subtract (rotatz in the clockwise direction) from Or an angle equal to 2pz' = 4Rz1/ik This will locate the point for lTIeJ\ which determines z , , the normalized input iinpedance looking into a lossless line of characteristic impedance R,, length z', and a nofmalized load impedance z,. Two additional scales in Az'li. are usually provided along the perimeterof the Irl = 1 circle for easy reading of the phase change 2p(Ar1) due to a change in line length A:': the outer scale 1s marked "wavelengths tow&-d generator" in the clockwise direction (increasing 2'); and the inner scale is marked "wavelengths toward l o a d in the counterclockwlse directlon (decreasing z'). Eigure 9-15 is a typical Smith chart, which is commercially available.' It has a complicatid appearance, but actually it consists merely of constant-r and constant-x circles. We note that a change ofhalf-a-wavelength In line length (A:' = i /2 ) corresponds?~ a i f l (Azl) = 2n change in 4. A complete revolution around a /TI-circle rettlrns to thesame point and results in no change in impedance, as was asserted in Eq. (9-95).

' All of the Srni!h charts used in th~s book are ~eprinted wlth perrnlsslon of Ernelold Industries, Inc., New Jersey.

416 THEORY AND APPLICATIONS OF TRANSMISSION LINES 1 9 ! . I

Fig. 9-15 The Smith chart. i i

I In the following we shall illustrate the use of the Smith chart for solving some j

typical transmission-line problems by several examples. 1 i r

d + $ Use the Smith chart to find the input impedance of a section of a I 50-(Q) lossless transmission line which is 0.1. wavelength long and is terminated in F b

a short-circuit. i

1. Enter the Smith chart at the intersection of r = 0 and x = 0 (Point P, on the extreme left of thart. See Fig. 9-16.)

2. Move along the pehmeter of the chart (1'1 = 1) by 0.1 "wavelengths toward generator" in a clockwise direction to P , ,

3. At P, read r = 0 and x 2 0.725, or z, = jO.725. Thus, Zi = Rozi = 50(~0.725) = j36.3 (0). (The input impedance is purely inductive.)

This result can be checked readily by using Eq. (9-92).

Z, = jR,, tan /I/ = j5O tan

= j5O tan 36 = j36.4 (Q).

A lossless transmlssion line of length 0.4341 and characteristic impedance 100 (11) is terminated in an impedance 260 + j180 (Q). Find (a) the vohape reflection coefficient, (b) the standing-wave ratio, (c) rhe input impedance, and (d) the location of a voltage maximum on the line. .

.I .. Solution: Given .

'e find tlie voitage reflection coefficient in several steps:

Enter the Smith chart at r, = Z L I R o = 2.6 + jl.8 (Point P , in Fig 9-16.) With the cedtw at the origin, draw a circle of radius W, = T J = 0.60. (The radids of the chart OFsc equals unity.) Draw ,~e.straight line OP, and extend it to P2 on the periphery. Read 0.220 on "wavelen&hs towardgenerator" scale. The phase angle 6, of the reflectmn coeffi~ient is (0.250 - 0.220) x 4n = 0 . 1 2 ~ (rad) or 21'. (We multlply the change in wavelengths by 4n because angles on the Smlth chart are measured in 2p?c: in.-'/I,. A hiill-w:lveleogth change in line lengtll corresponds :o a complck ~aro~uiion on the Suit11 chart.) The answer to part (a) is then

r = jrleJ8>= 0.60/21".

h) The (rl = 0.60 circle intersects wit6 the positive-real axis OP.. at r = S = 4. Thus the voltage s twndi~~~-wavc ratio is 4.

, ,418 THEORY AND APPLICATIONS OF TRANSMISSION LINES / 9 * .

Fig. 9-16 Smith-chart calculations for Examples 9-9 and 9-10.

c) To find the input impedance, we proceed as follows:

1. Move P; at 0.220 by a total of 0.434 "wavelengths toward generator," first to 0.500 (same as 0.000) and then further to 0.1 %[(OSOO - 0.220) + 0.1 54 = 0.4341 to P;.

I .- 2. Join 0 and P; by a straight line which intcrsects the Irl = 0.60 circle at P,.

rcle at P,.

., L ,

;r 1 '1 i 9-5 / THE SMITH CHART 419

f 1 ,

3. Read r = 0.64 and x = 1.2 at P,. Hencd, . . . ; z;= ROz, = lOO(q69 + j1.2) = 69 + j120 (R).

d) In going from Pi to P,, the lrl = 0.60 circle intersects the positive-real axis OPoc at P, where thd voltage is a maximum. Thus, a voltage maximum appears at (0.250 - 0.220)A or'0.030A from the load.

Example 9-11 ~ o h e E~ample 9-7 by using the Smith chart. Given

4 Ro .? 50 (Q)

w S=3.0 1,. ,, 2 = 2 x 0.2 = 0.4 (m) I First voltage minimum at rk = 0.05 (m),

find (a) a), (b) Z,, (c)'&, 2nd R,,, (Fig. 9-1 1).

Solution

a) On the positive-rcill dris OP,. locale ~ h c po~nt I>,,, .lr wlilcb = s = 3.0 (see F , ~ . 9-17). Then DM = /r/ = 0.5 (mot = 1.0). We cannot find 0, until we have located the point that represents the normalized load imped;ince.

b) We use the following procedure to find the load ihpedance on the Smith &art:

1. Draw a circle centered at the origin with radius m,\,, which intersects with the negative-r a1 anis OP,. at P,,, where there will be a voltage minimum.

2 Since &/i = J05/0.4 = 0.125, move from P,. 0.125 '6wav&ngths toward load7' -

in the counterclosewise direction to P;. 3. Join 0 and P; by a straight line, intersecting the (TI = O.j circle at P,. This

1s the point rebrescnling the normalized load impedance. 4. Read the angle i POCOP; = 90' = n/2 (rad). There is no need to use a pro-

tractor, because L POCOP; = 4n(0.250 - 0.125) = n/2. Hence 6 , = - @ (nd) , or r = 0.51-40" = - jO. j .

5- Read at PL, 2, = 0.60 - j0.80, which gives .

All the above results are the same as those obtained in Example 9-7, but no calculations with Complex numben are needed in using the Smith chart.


Fig. 9-17 Smith-chart calculations for Example 9-11.

9-5.1 Smith-Chart Calculations for Lossy Lines

In discussing the use of the Smith chart for transmission-line calculations, we hav.e assumed the line to be lossless. This is normally a satisfactory approximation for we generally deal with relatively short sectioxis of low-loss lines. The lossless assumption enablcs us to say, following Eq. (9-130), that the magnitude of the Te-12"' term

S' ? 1 I does not change with line length z' and that we can find z, from z,, and vice versa,

1 1 I 8

by moving along the lrikircle by an angle eqhal to 2p.2'. I

For a lossy line of a sufficient length 8, sdch that 2 d is not negligible compared i to unity, Eq. (9-136) must be amended to read

. r

1 + re - 2 a ~ ' ~ - j2pz'

2, = 1 - re- 2az' -12pr' e

' l + ( r ( e - 2 a - ' e ~ 9

- - 1 - 6 = or - 2 w . (9-132)

,r Hence, to find I , @om z,, we cannot aimply move along the (TI-circle; auxiliary calciilations are necessary in order to account for the e-'"' factor. The following example illustrates What has ta be done.

. we nave on for we sumption 12i7:' 4 term

Example 9-12 Tlle input impedance of n short-circuited loshy transmission line of length 2 (m) and characteristic impedance 75 $2) (approximately real) is 45 + j225 (0). (a) Find a and /3 of the line. (b) Determine the input impedance i f the short-circuit is replaced by a lo:ld impc~lilncc Z , = 67.5 - j45 (Q).

Solution

a) The short-circuit load is represented by the point P, on the extreme left of the Smith impedance chart.

1. Enter zil = (45 + j225)/75 = 0.60 + j3.O in the'chart as P , (Fig. 9-18), 2. Draw a straight line from the origin OthrouSh P , to P;. -- 3. Measure OP,/OP', = 0.89 = c-". It follows that

4. Record that the arc P,P; is 0.20 "wavelengths toward generator." We have //A = 0.20 and 286 = 4n//A = 0 . 8 ~ . Thus,

b) To find the input impedance for Z , = 67.5 - j45 (R): -

I . h n k r ZL = %,/2, = (67.5 - /45)/75 = 0.9 - jO.6 on thc Smith chart as P,. 2. Draw a straight line from 0 through P, to P2 where the "wavelengths toward

generator" reading is 0.364. - 3. Drawa/r(-circle centered at 0 with radius m2. 4 Move Pi alohg the perimeter by 0.20 "wavelengths toward generator.' to

P i at 0.364 + 0.20 = 0.564 or 0.064. , 5. Join P3 and d by a straight line, intersecting the /r/-circle at P,.

6. Mark on line OP, a point Pi such that 0P1/CF3 = e-2a' = 0.89. 7. At Pi, read zi * 0.64 + j0.27. Hence,

2, = 75(0.64 + j0.271 = 48.0 + j20.3 (R).

\ 422 THEORY AND APPLlcATloNs OF TRANsMlssloN LINES 1 9 e

Fig. 9-18 Smith-chart calculations for lossy transmission line (Example 9-12).

9-6 TRANSMISSION-LINE IMPEDANCE MATCHING

Transmission lines are used for the transmission of power and information. For radio-frequency power transmission it is highly desirable that as much power as possible is transmitted from the generator to the load and as little power as possible is lost on the line itself. This will require that the loztd be matched to the characteristic

i

I

9-6.1 Quarter

/.

> I :

9-6.1 Impedance ~ a k h l h ~ by Quarter-Wave Transformer '

Kb = ,/=. (9-133)

Since the length of the quarter-wave line depends on wavelengrh, this matching is frequency-~eniitive. as are a]! the other methods to be discussed.

Example 9-13 A &pal generator is to feid equal power through a lossless air transmission line with a characteristic impedance 50 (Q) to two separate resistive loads, 64 and 23 (n). Quarter-wave transformers are used to match the ]oa& to the 50 cn) line, as shouo in Fig. 9 -1% (a) Determine the required charactenstic impedances of the quarler-wave lines. (b) Find the standing-wave ratios on the matching line scctiohs. , .

fig. 9-19 impedance matching by quarter-wave lines (Ex:irnplr 9- 13).

i


Solution

a) To feed equal power to the two loads, the input rehstance at the junction with the main line' looking toward each load must be equal to 2Ro. Ri l = Ri2 = 2Ro = 100 (Q).

~b~ = J ~ = J i W Z i = 8 o ( n )

G2 = Jm = 4TGE-Z = 50 (Q).

b) Under matched conditions, there are no standing waves on the main transmission line (S = 1). The standing-wave ratios on the two matching line sections are

Matching section No. 1 .

Matching section No. 2

Ordinarily the main transmission line and the matching line sections are essentially lossless. In that case both Ro and Ro are purely real and Eq. (9-133) will have no solution if R L is replaced by a complex ZL. Hence quarter-wave transformers are not useful for matching a complex load impedance to a low-loss line.

In the following subsection we will discuss a method for matching an arbitrary load impedance to a line by using a single open- or short-circuited line section (a single stub) in parallel with the main line and at an appropriate distance from the load. Since it is more convenient to use admittances instead of impedances for parallel connections, we first examine how the Smith chart can be used to make admittance

. calculations. Let YL = l/ZL denote the load admittance. The normalized load impedance is

where YL = YLP0 = YLIGo .

= ROYL = g + j b (Dimensionless), (9-135)

ion with = Rtz =

smission .S are

re essen- will have mers are

, ubitrary sction (a from the r'parallel mittance

is the normalized load admittance having normalized conductance g and normalized susceptance h as its teal and imagin;~ry pdrts respectively Equ;ltion (9-134) suggcsts 111:11 ; I qtmrlcr-w;tvt l i :w will, ;I t t~l~ly ilornlilli~d char:~~tert~t ic impedance will l j ' m h - r n 2 , lo y,, i l t iU vice versa. On the Smith chart we need only to move the

. point representing T L along the IT(-circle by a quarter-wavelength in order to locate thc point r cp re s~n t ih~ ,. Since a 214-change in line length (AzlN = i) corresponds to a change of n radians (2PAzf = x) on the Smith chart. the poir~ts rrpreset~titlg Z,

and y, are then diametrically opposite to eachuther on the Jr/-circle. This observation enables US to find yi. from r,, and z, from y,, on the Smith chart in a very simple manner.

Solution: This problem has nothing to do with any transmission line. In order to use the Smith chart, we can choose an arbitrary normalizing constant; for instance, RO = 50 (R). Thus,

Enter z, as point P, on the smith chart in Fig. 9-20. The point P, on the other side of the line joining PI and 0 represents y,: 07, = m1.

-1-

Example 9-15 ~ i n d the input admittance of an open-circuited line of characteristic impedance 300 (a) and leapth 0.042.


a-

Fig. 9-21 Finding input admittance of open-circuited line (Example 9-15).

Solution

1. For an open-circuited line, we start from the point Po, 6nYthcextreme right of the impedance Smith chart, at 0.25 in Fig. 9-21.

2. Move along the perimeter of the chart by 0.04 "wavelengths toward generator" to P 3 (at 0.29).

3. Draw a straight line from P3 through 0, intersecting at P; on the opposite side.

4. Read at P; y, = 0 + jO.26.

Thus, 1 x=-

300 (0 + j0.26) = j0.87 (mS).

In the preceding two examples we have made admittance calculations by using the Smith chart as an impedance chart. The Smith chart can also be used as an admittance chart, in which case the r and x circles would be g and h circles. The points representing an open- rind short-circuit termination would be the points on the extreme left and t%e extreme right, respectively, on an admittance chart. For Example 9-15, we could then start from extreme left point on the chart, at 0.00 in Fig. 9-21, and move 0.04 "wavelengths toward generator" to P; directly.

9-6.2 Single-Stub Matching

We now tackle the problem of matching a load impedance 2, to a lossless line that has a characteristic impedance R, by placing a single short-circuited stub in parallel with the line, as shown in Fig. 9-22. This is the single-stub method for impedance matching. We need to determine the length of the stub, L', and the distance from thc load, z', such that the impedance of the pariillel combination to the right of points B-B' equals R,. Short-circuited stubs are usually used in preference to open-circuited

ttance 9-15).

I rght of

r' eratc

tte stae.

y using i as an es. The 1111h fill

l l 1'111

0.W i n

n ne th, 3ara" '

ledan, om the points

rcuited

$ 2 I

stubs ~ C C ; I U S C i l n inlinitc tormin:~ling i m a d i ~ n u : IS more difficult to re;~lim (hiin il zero t o d i n a t h i irdpcdancc for reasons of radiation from an open end and coupling effects with neighbbring,objects. Moreover, a short-circuited stub of an adjustable length and a constant qharacteristic resistdhce is much easier to construct than an open-circuited one. Of course, the difference in the required length for an open- circuited stub and,that for a short-circuited stub is an odd multiple of a quarter- wavclcncth. -

The parallel cdhbination of a line terminated in Z, and a stub at points B-B in Fig. 9-22 sugsddts that it is advantageous to analyze the matching requirements in terms of admitt~hces. The basic requirement is

> -

In terms of normalized Lidmittances. Eq. (9- 136) becomes

1 = + J*,, (9-137)

where y , = ROY, is for tne load sectloll and y , = ROY, is for the short-circurted stub. However, since thdi input admtttance of a short-circuited stub is purely susceptive. y, is purely imaginary. As a consequencg.Eq. (9-137) can be satisfied only if

and YII = 1 + jb, . (9-138a)

Y B = - J ~ B , (9-138b) where b, can be either positive or negative. Our objectives, then, are to find the length d such that the admittance, y,, of the load section looking to the rlght of terminals B-E' has a hnity r e d purt and to find the length fB of the stub required to cancel the imuginury purr.

Fig: 9-22 Impedance matching Y by ringicrtvb method.

428 THEORY AND APPLICATIONS OF'TRANSMISSION~LINES / 9

Using the Smith chart as an admittance chart, we proceed as follows for single- , t

stub matching:

1. Enter the point representing the normalized load admittance y,.

2. Draw the Irl-circle for y,, which will intersect the g = 1 circle at two points. At these points y,, = 1 + jbBl and y,, = 1 + jb,,. Both are possible solutions.

3. Determine load-section lengths dl and d2 from the angles between the point representing y, and the points representing y,, andly,,.

4. Determine stub lengths t,, and tB2 from the angles between the short-circuit point on the extreme right of the chart to the points representing -jb,, and - jh,], respcctively.

The following example will illustrate the necessary steps.

Example 9-16 A 50-(R) transmission line is connected to a load impedaice Z, = 35 - j47.5 (Q). Find the position and length of a short-circuitcd stub rcquircd to march rhc linc.

\--

Solution: Given R o = 50(Q) ., 2, = 35 - j47.5 (R) z, = Z,/R, = 0.70 - j0.95.

1. Enter z, on the Smith chart as P , (Fig. 9-23). 2.. Draw a Irl-circle centered at 0 with radius m,. 3. Draw a straight line from P, through 0 to point P i on the perimeter, intersecting

the Il-1-circle at P,, which represents y,. Note 0.109 at P; on the "wavelengths toward generator" scale.

4. Note the two points of intersection of the Irl-circle with the g = 1 circle.

At P,: y,, = 1 + j1.2 = 1 + jb,,; AtP,: y,,= 1 -j1.2= 1 +jbBz.

5. Solutions for the position 'of the stub:

For P, (from P; to P;): dl = (0.168 - 0.109)A = 0.0591.; For P4 (from P; to f i ) : d, = (0.332 - 0.109)A = 0.2231.

, 6. Solutions for the length of short-circuited stub to provide y, = -jb,: For P , (from P,, on the extreme right of chart to P;', which represents -jh,, = -j1.2):

For P4 (from P,, to Pi, which represents -jb,, = j1.2):

r single-

t . .. < t i

points. : - Jutions.

e point

sccting lengths

Fig. 9-23 Construction for single-stub matching.

In general,. the solution with the shorter lengths is preferred unless there are other practical constraints. The exact length, t'B, of the short-circuited stub mey require fine adjustments in the actual matching procedure; hence the shorted matching sections are sometimes called stub tuners.

The use of Smith chart in solving impedance-matching problems avoids the manipulation of complex numbers and the computation of tangent and arc-tangent

.?. _. I

430 THEORY AND APPLICATIONS OF TRANSMISSIQN LINES / 9 1

functions; but graphical c~nstructions are needed, and graphical methods have limited accuracy. Actually the analytical solutions of jmpedance-matching problems are relatively simple, and ebsy access to a computer'may diminish the reliance on the Smith chart and, at the same time, yield more accurate results.

For the single-stub matching problem illustrated in Fig. 9-22, we have, from Eq. (9-89),

where t = tanpd. (9 - 140)

The normalized input admittance to the right of points B-B' is

where

and

A perfect match requires the simultaneous satisfarion of Eqs. (9-138a) and (9-l3Sb). Equating g, in Eq. (9-142a) to unity, we lii~vc .

(r: - I)t2 - 2sLt + (I., - 1.; - .yl) = 0. (9-133)

Solving Eq. (9-143), we obtqin

The required length d can be found from Eqs. (?-140), (9-144a), and (9-1 44b):

- tan-'t, t 2 0 ,

- (n + tan-It), t < 0 .

Similarly, from Eqs. (9-138b) and (9-142b), we obtain

ds have roblems ance on

re, from

(9-139)

(9-140)

(9-141)

1 - l32a)

/--

14%)

I %b),

9-143)

- 1 Ma)

-144b) I

4b): Y

- 145a)

- 145b) /I

- 1464,-

-146b)

1 , ~, - For a given load [mpdi\nw, both ,112 i tn?;f/~ can bc determined easily on a scientific calculato~. It$ also a simple matter t a write a general computer program for the single-stub mhtchidg problem. Mbre &curate answers to the problem in Example 9-16 (r, = 0170 and s, = -0.95) are

Of course, such accuracies are seldom needkd in dn actual problem; but these answers have been obtained easily without a Smithkhart,

rl ' . Double-Stub ~a tch ing

The method of impedance matching by mean8 of a single stub described in the preceding subsection can be used to match any atbitrary, nonzero. finite load impedance to the characteristic resistance of a line However, the single-stub method requires that the stub be nrtnchcd to the main lidc at a specific point which varies as thc loild impedance is changed. This requirement &en presents practical difficulties because the specified junction point may occur at an undesirable location from a mechanical viewpoint, Furthcrmorc. it is very dificult to build :1 variable-lengh C L X I X ~ ~ I ~ linc will1 ;I conslu~~t:ch;m~clcristic impcdilncc. In such c;~scs, ttn altcmatire mclhod for irnpcd;~nc~-m;ltcbing is to us, two short-ciccuitcd stubs i~ttilchcd to the main line a1 lined posilion:.., i s shown in Yig.'9:24. Here, the distance do is fined and arbitrarily chosen (such as L/16. JL/8, 3 / : /16 ,3 i /8 , etc.), and the lengths of the two stub tuners are adjusted to match a given load impedahce Z , to the main line. This scheme is the double-.stub melhid for impedanm matchink.

In the arrangement in Fig. 9-24, a stub of length e, is connected directly in parallel with the load itnped&nce Z , at terminals A-A', and a second stub of length

I

Fig. 9-24 Impedance matching by double-stub method.

I 432 THEORY~AND APPLICATIONS OF TRANSMISSION LINES 1 9

lB is attached at terminals B-B' at a fixed distance do away. For impedance matching I w i t h a main line that has a characteristic resistance R,, we demand the total input

admittance at terminals B-B', looking toward the load, to equal the characteristic conductance of the line; that is,

In terms of normalized admittances, Eq. (9-147) becomes

Now, since the input admittance y,, of a short-circuited stub is purely imaginary, Eq. (9-148) can be satisfied only'if

.v,, = I + jh, (9 - 140:1) and

Note that these requirements are exactly the same as those for single-stub matching. On the Smith admittance chart, the point representing y, must lie on the g = 1

circle. This requirement must be translated by a distance d,/L "wavelengths toward load"; that is, y, at terminals A-A' must lie on the g = 1 circle rotated by an angle 4xd,/L in the counterclockwise direction. Again, since the input admittance y,, of the short-circuited stub is purely imaginary, the real part of y, must be solely con- tributed by the real part of the normalized load admittance, g,. The solution (or solutions) of the double-stub matching problem is then determined by the intersection (or intersections) of the g,-circle with the rotated g = 1 circle. The procedure for solving a dou'ble-stub matching problem on the Smith admittance chart is as follows.

1. Draw the g = 1 circle. This is where the point representing y, should be located.

2. Draw this circle rotated in the counterclockwise direction by d,JA "wavelengths toward load." This is where the point representing y, should be located.

3. Enter the y, = g, + jb, point.

4. Draw the g = g, circle, intersecting the rotated g = 1 circle at one or two points where y, = g~ + jbA.

. ..." 5. Mark the corresponding y,-points on the g = 1 circle: y, = I f jb,. 6. Determine stub length t', from the angle between the point representing y, and

the point representing y,. 7. Determine stub length t', from the angle between the point representing -jbB

and P,, on the extreme right.

e matching , 1 l i

total input f . - aracteri~tic

, , 1; I ' J 1 '

i

imaginary,

I , l~l,ltcllill&

, I ? ' , = I

t'lncc t,, of wlely con-

iolution (or y the mter- IZ procedure

! chart is as

1 be located.

[:lo points

/Itinr and '

enting -jb,

.' t

line is connected to a load impedance Z , = i 60 + j80 (a). A an e&hth of a wavelength apart is used to

9-24. Find the required lengths of the

L 1, < ' ? I

Solution: Given R, = 50 (R) and Z, = a + 180 (R), it is easy to calculate

t' (We could find y~ on the Smith chart by locating the point diametrically opposite to z , = (60 + j80)/50 = 1.40 + j1.60, but this would*clutter up the chart too much.) We follow the procedure outlined above. '

1. Draw the g = 1 dircle (Fig. 9-25).

circle by $ "wavelengths toward load" in the counterclockwise of rotation is 4x18 (rad) or 90'.

3. Enter y , = 0.30 - j O 40 as P,.

4. Mark the two pgints of intersection, PA, and P A , , of the g, = 0.30 circle w ~ t h Ihc rol;ltctl (1 -- I c1rc.1~.

At P A , , read y,, = 0.30 i jQ.29 ;

At PA,, ready,, = 0.30 + jl.75.

5. Use a com$stss ccntcrbd at the origin 0 to mark the points P,, and P,, on the g = 1 circle correspoqding, respectively, to the points P A , and PA,.

At P,,, read y,, = 1 + jl.38;

, At P,,, read y,, = i - j3.4. 5 )

6. Determine the reguiri;d stub lengths t,, axid tA2 from \

(.Vsn)l =1',,, -j1,=j0.69, =(0.097+0.250)%=0.3471.(PointA,), (y;,), * yd2 L YL = j2.11, tA2 = (0.17'9 + 0.250)/1 = 0.4291, (Point A,).

2 .' '

7. Determini-the required stub lengths /,, and fB2 from:

(&)I = -jl 38, Ll = (0.350 - 0.250)iL = O.lOOit (Point B,) , a -

-(jj;;J2 = ~ , 2 . 4 , f,{, = (0.205 t. 0.3O)i. = 0.4551. (Point B?).

~ x a m i d i i b t l 0; {he ~cns t rwt ion in Fig. 9-25 reveals that if the point P,, representing the n&~r@liied lead a h i t t a n c e yt = g, + jb, lies within the g = 2 circle (if gL > 2), then t& 3 = gi circle does notjntersect with the rotated g = 1 circle and no solution exists for double-stub matching with do = R/8. This region for no solution varies with the chosen distanced, between the stubs (Problem P.9-38). In such cases

Fig. 9-25 Construction for double-s~ub matching.

impedance matching by the double-stub method can be achieved by adding an appropriate line section between Z, and terminals A-A', as illustrated in Fig. 9-26 (Problem P.9-37).

An analytical solution of the double-stub impedance matching problem is, of course, also possible, albeit more involved than that of the single-stub problem

P

diiing an Fig. . '6 .

-

lem is, of problem

I Fig. 9-26 ~ouble -s thb irlpedhnce matching with added load-line section.

developed in the preceding subsection. The more ambitious reader may wish to obtain such an analytical solution and write a computer program for determining dJA /*/A, and l B / A in terms of z, and dJ i .

REVIEW QUESTIONS

R.9-1 Discuss the similarities and dissimilarities of uniform plane waves in an unbounded media and TEM waves along transmission lines.

H.9-2 What are the Ihree diost common typesof guiding structures that support TEM wava?

R.9-3 Compare the advantages and disadvantages of coaxial cables and two-wire transmission lines.

R-9-4 Write the t rad~mk~on- l ine equations for a lossless parallel-plate line aupport~ng TEM waves.

R.9-5 What are str~phnes ?

R.9-6 Describe how the character~stm Impedance of a parallel-plate transmiss~on l ~ n c depcnda on plate w~dth and dielectric thickni~s.

H.9-7 C w ~ p ; ~ r c lhv ~ e l u r ~ l y 01' TEM-wave propagation along a parallel-plate transmasion line with that i32n unbounded medium. . R.9-8 D c f i n ~ ,q,hce imp&nce. How is surface itnpedance related to the power dissipated i,n a plate condpcror?

R.9-13 State the differtnoe between the surfac6 resistance and the resistance per u n ~ t length of a parallel-plate transmislon line. .

>

, . 436 . THEORY AND APPLICATIONS OF TRANSMISSION LINES 1 9

R.9-11 What is the essential dikerence between a transmission line and an ordinary electric network?

R.9-12 Explain why waves along a lossy transmission line cannot be purely TEM.

R.9-13 Write the general transmission-line equations for arbitrary time dependence and for time-harmonic time dependence.

R.9-14 Define propagation constant and characteristic impedance of a transmission line. Write their general expressions in terms of R, L, G, and C for sinusoidal excitation.

R.9-15 What is the phase relationship between the voltage and current waves on an infinitely long transmission line?

R.9-16 What is meant b; a "distortionless line"? What relation must the distributed parameters of a line satisfy in order for the line to be distortionlcss?

R.9-17 Outline the procedure for'determining the distributed parameters of a transmission line.

R.9-20 On wh;~t factors does the input i~npcdance fa transmission line depend'!

,A.

R.9-21 What is the input impedance of an open-circuited lossless transmission line if the length of the line is (a) %/4, (b) 1/2, and (c) 3i./4?

R.9-22 What is the input impedance of a short-circuited lossless transmission line if the length of the line is (a) A/4, (b) 142, and (c) 31./4?

R.9-23. Is the input reactance of a transmission line 1/8 long inductive or capacitive if it is (a) open-circuited, and (b) short-circuited?

R.9-24 On a line of length C, what is the relation between the line's characteristic impedance and propagation constant and its open- and short-circuit input impedances?

R.9-25 What is a "quarter-wave transformer"? Why is it not useful for matching a cornplcx load impedance to a low-loss line?

R.9-26 What is the input impedance bf a lossless transmission line of length C that is terminated in a load impedance Z, if (aJ C = 142, and (b) f = R?

R.9-27 Define voltage reflection coeficient. Is it the same as "current reflection coetEcien!"'? Explain.

R.9-28 Define standing-wave ratio. How is it related to voltage and current reflection coefficients?

R.9-29 What are r and S for a line with an open-circuit tcrmination? A short-circuit tcrmination?

R.9-30 Where do the minima of the voltage standing wave on a lossless line with a resistive termination occur (a) if RL > R,, and (b) if R, < R,?

* R.9-31 Explain how the value of a terminating resis!ancc can be determined by measuring the standing-wave ratio on a lossless transmission line.

. Write

finitely

meters '

sdance

:ient'"!

:]ems?

a t i o n 7 p

Gstive

- ng the

H.9-32 Explain value of an arbitrary t e m i d t i n g impedance on a lossless transmission standirig-wave measbIemedh on the line.

R.9-33 A voltage genekitor having an internal impe&me 2. is connected at I = 0 to the input terminals of a lossless t dsmigsion line of length t . h e line has a characteristic impedance Z, and is terminated with i ~ p e d a n c e Z,. At what t h e will a steady state on the line be reached i f (a) Z,, = 2, and Z, = e,, (b) Z, = Z, but Z, # Z,, (c) Z, = Z, but 2, # Z,, and (d) Z, # 2, and Z, # Z,?

I

R.9-34 What is a smith chad>and why is it u6eful id making transmission-line calculations?

! R.9-35 Wlfere is the pb nt rep'resking a matchcd lodd on a Smith chart?

, R.9-36 For a given loaa impdance Z , on a lossless l k e of characteristic impedance Z,, how do we use a Smith chart to, hetermine (a) the reflect& coefficient anti (b) the standing-wave ratlo?

R.9-37 Why does a chdnge of half-a-wavelength in line length correspond to a complete rcvo- lution on o Smith chart I R.9-38 Given an impedance Z = k + jX, what procedure do we follow to find the admittance Y = 1/Z on a Smith chart?

K.9-39 Given an admittance Y = G + JB, how d o we use a Smith chart to find the impedance z = I I Y ?

R.9-40 Where is the poifit representing a short;circuit on a Smith admittance chart?

R.9-41 Is the standing-have xtio constant on a trdd9mission.line even when the line is lossy? Explain.

R.9-42 Can a Smith c h t b: used for impedance ccilculations on a lossy transmlsslon line'! Explain.

R.9-43 Why is it more';convenient to use a Smith chart as an admittance chart for solving impedance-matching problems than to use it as an impedance chart?

R.9-44 Explain the single-stu!, method for impedance matching on a tranarn~ssion line.

1t.O-45 Bxpliiin lhc doublc-sit~h rncthod for impcdancc matching on a transmission Line.

R.9-46 Compare the relative ddvantages and disadvantages of the single-stub and the double- stub methods of impedance matching.

PROBLEMS

P.9-1 Neglecting fringe fields, prove analytically that B y-polarized TEM wave that propagates along a parall&piate tratlsmission line in + z direction has the following properties: dE,/dx = 0 and dH,/dy = 0.

P.9-2 The electric and hagnetic fields of a gcneral TEM wave traveling in the + z direction alohg a transmission line may have both x and y components, and both components may be functions of the transverse dimensions. . !

a) Find the relatioh amang EJx. y), E,(s, y), H,,(.t, y), and H,(s, y).

438 THEORY AND APPLICATIONS OF TRANSMISSION LINES / 9

b) Verify that all the four field components in part (a) satisfy the two-dimensional Laplace's equation for static fields.

P.9-3 Consider lossless stripline designs for a given characteristic impedance.

a) How shodld the dielectric thickness, d, be changed for a given plate width, w, if the dielectric constant, E,, is doubled?

b) How should w be changed for a given d if c, is doubled? c) How should w be changed for a given E, if d is doubled? d) Will the velocity of propagation remain the same as that for the original line after the

changes specified in parts (a), (b), and (c)? Explain.

P.9-4 Consider a transmission line made of two parallel brass strips:-a, = 1.6 x 10' (S/m)-- of width 20 (mm) and separated by a lossy dielectric slab-p = p,, r , = 3, a = (S/m)-of thickness 2.5 (mm). The operating frequency is 500 MHz.

a) Calculate the R, L, G, and C p& unit length. 1)) Compnrc the magnitudes of thc nxinl ant1 Ir;~nsvcrsc components of the electric field. c) Find 7 and Z,, .

P.9-5 Verify Eq. (9 -39). -1 -

P.9-6 Show that the attenuation and phase constants for a transmission line with perfect' conductors separated by a lossy dielectric tlii~t.,llas u complex prnmittivity E = c' - jc" arc, resncctivclv.

P.9-7 In the derivation of the approximate formulas of y and 2, for low-loss lines in Subsection 9-3.1, all terms containing the second and higher powers of (RIoL) and (G/wC) were neglected in comparison with unity. At lower frequencies better approximations than those given in Eqs. (9-45) and (9-47) may be required. Find new formulas for y and Z , for low-loss lines that retain terms containing (RIwL)' and (GIwC)'. Obtain the corresponding expression for phase velocity.

P.9-8 Obtain approximate expressions for y and Z, for a lossy transmission line at very low frequencies such that wL << R and o C << G.

P.9-9 The following characteristics have been measured on a lossy transmission line at 100 MHz:

Determine R, L, G, and C for the line. \ P.9-10 It is desired to construct uniform transmission lines using polyethylene (E, = 2.25) as the dielectric medium. Assuming negligible losses, (a) find the distance of separatiqn for a 300-(R), two-wire line, where the radius of'the conducting wires is 0.6 (mm); and (b) findthe inner radius ofthe outer conductor for a 7542) coaxial line, where thc radius ofthc ccnter conductor is 0.G (mm).

ne after the

ctric ficld.

.,vitb perfect m : e ,

,n Subsection ere neglected glven in Eqs. es that retain hase velocity.

e at very low

eat 100 MHz:

- 2' s the for a >JO-(R),

ie inner radius tor is 0.6 (mm).

8 : 8

PROBLEMS 439 I . .' !

f P.9-11 Prove that a maximum power is ,transferred from a voltage source with an internal impedance Z, to aload impedance 2, over a~!ossle$s transmission line when 2, = 2:. What is the

,. , . maximum power-transfer efficiency?

I r -. 1 P.9-12 Express V ( Z ) and I(z) in terms of the voitage and current I, at the input end and y 'and Z , of a transmisdion line (a) in exponential fohn and (b) in hyperbolic form.

P.9-13 A DC generhtor o i voltage C, and internal resistance R, n connected to a lossy transmission line characterized by a resistance per unit length R and a conductance per unit length G.

a) Write the governing voltage and current ttansmission-line equations. b) Find the gedkral solutions for V ( z ) and I@). c) Specialize tH'e solutions in part (b) to those for an infinite line. d) Spccialim tht solutions in part (b) to thosc for a finite line of length L that 1s termmated

in a load redstance RL.

P.9-14 A gcncrator with n.n opcn-circuit volwge v,(t) = 10 sin XOOOnl (V ) and intcrnal impedance %,, .- 40 +-,j301!1) is conilclftcd 1 0 :I 50-(<1) ~ I i ~ ~ o i ~ ~ i c m l s ? ; ~ linc. 'l.hl: lir~c I W S :I rcslslancc of03 (f>/rn). and its lossy dielscLtk rncdi;tm has a loss tangent of 0.18%. The line is 50 (km) long and is tsrmi- nated in a matched load. Find (a) the instantaneous expressions for the voltage and current at an arbitrary location on the line, (b) the instantaneous expressions for the voltage and current at the load, and (c) the av&e power transmitted to the load.

P.9-15 The Input impedance of a n open- or short-circuited l o ~ y transmmion h e ha5 both d

resistwe and, a reactlve component. Prove that the input mpedance of a very short section C of a shghtly lossy line (a4 c 1 and P/ << I ) 1s approximately '

a) Z,, =i (R + fd)/ w ~ t h a short-c~rcu~t teimlnat~on. b) Z,, = (G - I~c) / [G* + ( U C ) ~ ] ~ ' with an open-circuit t e n n a t i o n

P.9-16 A 2-(m) lossfess transmission line having a characteristic impedance 50 (R) is termmated with an irnwdance 40 + j301R) at an operaticg frequency of 200 (MHz). Find the input impedance.

P.9-17 The open-circuit and short-circuit impedhnces measured at the input terminals of a transmissio$he 4 (m) long are, respectively, 250- ((R and 360/20" (R).

a) Det~finine Z,, a, and p of the line. b) Detebnine R, L, G: and C.

P.9-18 A lossless quarter-wave h e section of characteristic impedance R, is terminated with an indwtive load impedance ZL 5 RL + jXL.

a) ~ r o v g that.the input impedance is effectively a resistance Ri in parallel with a capacitive reactance X,. Determine R? and Xi in terms of R,, R,, and X,.

b) F i ~ d tcs -?.ti0 of the magnitude of the voltage at the input to that a t the load (volrage tr(i;i:>?:rt~ti~ti/~t~ r(~tio! l~ , , l / l~ , l ) in t c m s of 2, and Z L ,

P.9-19 A 75-(R) lossless lj& is terminated in a load impedance Z , = RL + jXL.

a) What must be the relation between RL and X, in order that the standing-wave ratio on !$e line be 31

b) Find X,, if dt = 150 (Q). c) Where does the voltage minimum &arest to the load occur on the line for part (b)?

440 THEORY AND APPLICATIONS OF TRANSMISSION LINES I 9

P.9-20 Consider a lossless transmission line. s . . . a) Determine the line's characteristic resistance so that it will have a minimum possible

standing-wave ratio for a load impedance 40 + j30 (Q). b) Find this minimum standing-wave ratio and the corresponding voltage reflection

coefficient. c) Find the location of the voltage minimum nearest to the load.

P9-21 A lossy transmission line with characteristic impedance Zo is terminated in an arbitrary load impedance ZL.

a) Express the standing-wave ratio S on the line in terms of Z0 and ZL. b) Find in terms of S and Zo the impedance looking toward the load at the location of a

voltage maximum. C) Find the impedance looking toward the load at a location of a voltage minimum.

P.9-23 The standing-wave ratio on a lossless 300-(R) transmission linet'bnhated in an unknown load impedance is 2.0, and the nearest voltage minimum is at a distance 0.31. from the load. Determine (a) the reflection coefficient l- of the load, (b) the unknown load impedance Z L , and (c) the equivalent length and terminating resistance of a line, such that the input impedance is equal to 2 , .

P.9-24 Obtain from Eq. (9-114) the formulas for finding the length t,,, and the terminating resistance R, of a lossless line having a characteristic impedance Ro such that the input impedance equals Z i = R, + j X i .

P.9-25 Obtain an analytical expression for the load impedance ZL connected to a line of characteristic impedance Zo in terms of standing-wave ratio $ and the distance, zui., of the voltage minimum closest to the load.

P.9-26 A sinusoidal voltage generator with V, = O . l p (V) and internal impedance Z , = Ro is connected to a lossless transmission line having a characteristic impedance Ro = 50 (R). The line is f meters long and is terminated in a load resistance R, = 25 (R). Find (a) v, I i , VL, and 1,; (b) the standing-wave ratio on the line: and (c) the average power delivered to the load. Compare the result in part (c) with the case where R, = 50 (R).

P.9-27 A sinusoidal voltage generator u, =, ll0sinwt (V) and internal impedance 2, = 50 (R) . is connected to a quarter-wave lossless line having a characteristic impedance Ro = 50 (R) that

is terminated in a purely reactive load ZL = j50 (R). a) Obtain voltage and current phasor expressions V(zl) and I(zl). b) Write the instantaneous voltage and current expressions u(zi, t) and i(z1, t). c) Obtain the instantaneous power and the average power delivered to the load.

P.9-28 The characteristic impedance of a given lossless transmission line is 75 (R). Use a Smith chart to find the input impedance a n 0 0 (MHz) of such a line that is (a) 1 (m) long and open- circuited, and (b) 0.8 (m) long and short-circuited. Then (c) determine the corresponding input admittances for the lines in parts (a) and (b).

ssible

:don

jitrary

n o f a

n.

a load crlstic

mown . Isa* L. 4 i n

indtlng edance

)f char- 4 oltage

= Ro is The line (b) the )are the

= so (n) (R) that

P

S m m d open- ~g lnput

? P.9-29 A load impedadce 30 + 110 (Q) is connected d(b a l~ssless transmission line of length 0.1011. and characteristic im&danke 50 (a). Use a Smith. chart to find (a) the standing-wave ratio, (b) the voltage re~ectlbn coefficient, (c) the jnput impedance, (d) the input admittance, and (e) the location of the $b~tagdm~nimum on thiline.$

~ 9 - 3 0 Repeat probl+ P.9%29 for a load i m p d a n k 30 - jlO (a).

P.9-31 In a laborator;, exp&ment conducted d;l a SO-($2) lossless transmlss~on hne ternmated in an unknown load Ypedance, it is found that the standing-wave ratio is 2.0. The successive voltage minima are 2$(cm) apatt and. the first minknum occurs at 5 (cm) from the load. Find (a) the load i m p d a n d , and (b) the reflection meffiEient of the load. (c) Where would the fint voltage minimum be Idtated'if the load were replaced by a short-circuit?

P.9-32 Thc input imdduncc of a short-circuited l o s e transmission line of length 1.5 (m) (ci/2) and characteristic impedance.100 (R) (approximately real) is 40 - j280 (R).

a) Find a: and /I of the line. b) Determine the input impedance if the short-circuit is replaced by a load impedance ZL = 50 + j50 (Q).

c) Find the input impedance of the short-circuifed line for a line leneth 0.15i.. - P.9-33 A dipole antenna having an i n p ~ ~ t i i l i~di ince of 73 (0) is fed by i 200-(MHz) source through a 3W-(n) t w o 4 r e trsnsmission line. Desidn a quarter-wave two-wire air line with a 2-(cm) spacing to matcg the antenna to the 300-(Q) line.

P.9-34 The single-stub melhod is used to match a load impedance 25 + j25 (Q) to a 50-(R) transmission line. I'

a) Find the required length and position of a ihort-circuited stub made of a section of the same 50-(R) linc. '

b) Repeat part (a) asstming the short-circuited stub is made of a section of a line that has a characteristic! impedance of 75 (Q). '

P.9-35 A load impedance can be hatched to a transmission line also by using a single stub placed in series with the load I t an appropriate location, as shown in Fig. 9-27. Assuming ZL = 25 + j25 (Q), Ro = 50 (R), and KO = 35 (0). find d and I required for matching.

Fig. 9-27 Impedancg hatching by a series stub.

P.9-36 The double-stub method is used to match a load impedance 100 + j103 (Q) to a losrless transmission line of characteristic impedance 300 ($2). The spacing between the stubs is 3218, with

I ', . " , ,,,one stub connected directly in parallel with the load. ~etennind the lengths of ihe siub tuners if

, (a) they are both short-circuited, and (b) if they are both opensircuited. "

.. 6

,)l - -* P9-37 If the load impedance in Problem P.9-36 is changed to 100 + j50 (Q), one discovers

that a perfect match using the double-stub method with do = 3118 and one stub connected directly across the load is not possible. However, the modified arrangement shown in Fig. 9-26 can be used to match this load with the line.

a) Find the minimum required additional line length dL. b) Find the required lengths of the short-circuited stub tuners, using the minimum dL found

L, . in part (a).

P9-38 The double-stub method shown in Fig. 9-24 cannot be used to match certain loads to a line with a given characteristic impedance. Determine the regions of load admittances on a Smith admittance chart for which the double-stub arrangement in Fig. 9-24 cannot lead to a match for do = A/16,1/4, 3118, and 71/16.

d, found

' I lbads to , ,

:ces on a ! 1, .ead to a

'.\ I

10-1 - INTRODUCTION

In the preceding cha&er we studied the characteristic properties of transverse slec- tromagnetic ( T E ~ ) waves guided by transmission lines. The TEM mode of guided waves is one in which the tlectric and magnetic fields are perpendicular to each other and both are transvkrfe tc. the direction of propagation along the guiding line. One of the salient properties i ~ f TEM waves guided by conducting lines of negligible resistance is that the velocity of propagation of a wave of any frequency is the same

. as that in an unbounded dielectric medium. This was pointed out in connection with Eq. (9d1) and was-reinforced by Eq. (9-55).

TEM waves, however, are not the only mode of guided waves that can propagate on transmission lines;. nor are the three types .of transmission lines (parallel-platz, two-wire, and coaxial) mentioned in Section 9-1 the only possible wave-guiding structures. As a mattcr of hct, wc scc from Eqs. (9-45a) and (9-49a) that thc Litteri- ualion constant resulting iiom thc linite conductivity of the lines increases with R. the resistance per unit line length, that, in turn, is proportional to f i in accordance with Tables 9-1 and 9-2. Hence the attenuation of TEM waves tends to increase monotonically with frequzncy and would be prohibitively high in the microwave range.

In this chapter we first present a general analysis of the characteristics of the waves propagating along uniform guiding structures. Waveguiding structures are

'

called waveguides, of which ,the three types of ttansmission lines are special cases. The basic governini equations will be examined. We will see that, in addition to transverse electromagrfetic (TEM) waves, which have no field components in rhe direction of propagation, both tru'nscerse r~tnqrzitic (TM) waves with a longitudinal electric-field cornponefit and tmr~sverse electric (TE) \caws with a longitndir.~! magnetic-tield.eomponent crin also exist. Both TM and TE modes have characterisric cutof l fi.cyuerf~?c.s. Waves of frcquencics below thc cutolT frequency of a particillar lnode cannot propagate, atid power and signal transmission at that mode is possible only for frequencies highe;. than the cutoff frequency. Thus, waveguides operating in TM and TE modes are like high-pass filters.

Also in this chapter we will reexamine the. field and wave characteristics of parallel-plate waveguides with emphasis on TM and TE modes and show that all

, .,. ti

transvers~~field components can be expressed in teims of~E,'(z being thk diredkn ' ., & . > , . , a , ,

, , 1

$ . < . of propagation) for TM waves, and in ternis 6f Hz fbr TE wave's. The attenuation f . . . . . . I . , . . . - 1 .

. . . . . b- t . .., - . ,.#-> . , ., constants resulting from imperfectly conddctihg prates will be detkmiined for TM , . and TE waves, and we will find that the attenuation cbnstant depends, in a com-

I ' I I '

plicated way, on the mode of the propagating wave, as well as on frequency. For some modes the attenuation may decrease as the frequency increases; for other modes, the attenuation may reach a minimum as the frequency exceeds the cutoff frequency by a certain amount.

I I Electromagnetic waves can propagate through hollow'metal pipes of an arbitrary cross section. Without electromagnetic theory it would not be possible to explain

'?. the properties of hollow waveguides. We will see that single-conductor waveguides cannot support TEM waves. We will examine in detail the fields. the current and charge distributions, and the propagation characteristics of rectangular waveguides. Both TM and TE modes will be discussed. An analysis of the propertie$ of circular waveguides requires a familiarity with Bessel functions as a consequence of manipulating Maxwell's equations III cylindrical coordinates. Circular w:~voguidcs will not be studied in this book. In millly applic:ltions wave psopagation in a rcctan- ,oular waveguide in the dominant (TE,,) mode is desirable because the electric field in the guide is polarized in a fixed direction.

Electromagnetic waves can also be guided by an open dielectric-slab waveguide. The fields are essentially confined within the dielectric region and decay rapidly away from the slab surface in the transverse plane. For this reason, the waves supported by a dielectric-slab waveguide are called sudtrce waces. Both TM and TE modes are possible. We will examine the field characteristics and cutoff frequencies of those surface waves.

At microwave frequencies, ordinary lumped-parameter elements (such as inductances and capacitances) connected by wires are no longer practical as resonant circuits because thc dimcnsions of thc clcrncnts would hove to bc cxtrcmcly small, because the resistance of the wire circuits becomes very high as a result of the skin effect, and because of radiation. All of these difficulties are alleviated if a hollow conducting box is used as a resonant device. Because the box is enclosed by conducting walls, electromagnetic fields are confined inside the box and no radiation can occur. Moreover, since the box walls provide large areas for current flow, losses are extremely small. Consequently, an enclosed conducting box can be a resonator of a very high Q. Such a box, which is essentially a segment of a waveguide with closed end faces, is

. called a cavity resonator. We will discuss the different mode patterns of the fields inside rectangular cavity resonators.

10-2 GENERAL WAVE BEHAVIORS ALONG. UNIFORM GUIDING STRUCTURES

In this section we examine some general chatactcristics for waves propagating along straight guiding structures with a uniform cross section. We will assume that the waves propagate in the +r direction with a propagation constant -J = ci + jB that

firection , 8 . '

is yet to be aetermined. For harmonic time dependence with an angular frequency o, muation the depende&e on r ahd t for all field components can be described by the exponential for TM . , L factor

: aco rn - \ , . n * i : I i - = e ( j ~ t - Y Z ) = -ee j(ut - BZ) (10-la) x y . For

1 . I AS an exa.mpl; for a. bosine reference we may write the instantaneous expression for

or other the E field as I ne cutoff i a E(x, y, z ; t ) = %[EO(x, y)e( jU ' -Y ' ) ] , (10-lb)

) explain veguides rent and \-

of circu-

veguides a rcctan- x i : field

n

:ve: -e. rapidly

ives sup- and TE

~quencies

:h as in- resonant

-:y small, :he skin

.I ho!low nduc~ing In occur. :s tremely -y high Q. ; faces, is :he fields

Ing along rhat the - j/3 that

where EO(x, y) is il two-dimensional vector phasor that depends only on the cross- sectional coordinated. The instantaneous expression for the H field can be written in a similar way. Hence. in using a phasor representation in cquations relating field quantities, we may replacc partial derivatives with respect to i and z simply by products witki ( j o ) and ( - 7 ) respectively; the common factor r'jw'-"' can be dropped:

We consider a 9tra.ght waveguide in the form of a dielectric-filled metal tube having an arbitrary no t s section and lying along the .- axis, as shown in Fig 10-1. According to Eqs. (7-8.6) and (7-87), the electric and magnetic field intensities in the charge-free diclcktr:~ region inside satisiy the following homogeneous vector l-lclmholtz's equations:

and V'E + k'E = 0

V2H + k" = 0, (10-2b) where E and H are three-dimensional vector phasors; and k is the wavenumber

r-- k = Wd,LLE. (10-3)

The three-dimensiona! Laplncian operator V2 may be broken illto two parts: 2 FuIu2 for the cross-sectional coordinates and Vf for the longitudinal coordinate.

For waveguides with a rectangular cross section, we use Cartesian coordinates:

= V.<,E + ;q2E.

Combination of Eqs. (lb-2a) and (10-4) gives

Fig. 10-1 A uniform waveguide with an arbitrary cross section.

. . . . . . . - .tl , . . I . . . . 1 -. . * , 446 WAVEGUIDES AND CAVITY RESONATORS / 10. *: -

.... Similarly, from Eq. (10-2b) we have

v$H + (yf + k 2 ) ~ = 0. (10-6)

We note that each of Eqs. (10-5) and (10-6) is really three second-order partial differential equations, one for each component of E and H. The exact solutibn of these component equations depends on the cross-sectional geometry and the boundary conditions that a particular field component must satisfy at conductor-dielectric interfaces. We note further that by writing V:+ for the transversal operator V:,, Eqs. (10-5) and (10-6) become the governing equations for waveguides with a circular cross section.

t Of course, the various components of E and H are not all independent, and it is not necessary to solve all six second-order partial differential equations for the six components of E and H. Let us edamine the interrelationships among the six componcnts in Ca rtcsian coordinates hy cspnnding thc two sourcc-free curl c tp t ions . Ecp. (7-851) atid (7 -8.517):

From V x E = -jo,uH: From V x-W-,jocE:

Note that partial derivatives with respect to z have been replaced by multiplications by ( - y ) . All the component field quantities in the equations above are phasors that depend only on x and y, the common c-'' factor for z-dependence having been omitted. By manipulating these equations, we can express the transverse field components H;. H.:, E.!, and E: in terms of the two longitudinal components Ep and H:. For instance. Eqs. (10-la) and (10-8b) can be combined to eliminate E; and obtain H: in terms of E: and HP. We have

wund- ~ C M ~ C

2 Vxp il,ith a

:d it is ihe six . \ com- J,

ltions.

stions .s tha: : been

cotn- and

:: and

where

The wave behavior in a waveguide can be analyzed by solving Eqs. (10-5) and (10-6) respectively for the longitudinal components, and HZ subject to the required boundary conditions, and then by using Eqs. {lo-9) through (10-12) to determine the other components.

It is convenient to classify the propagating waves in a uniform waveguide into three types according to whether E, or H , exists.

1. ~rnnsvers~'~lect~oma~netic ( T E M ) Waves. These are waves that contain neither Ez nor H;: We eehoontered TEM waves in Chapter 8 when we discussed plane waves and in Chlpter 9 on waves along tr&cimission lines.

2. Transverse Maglati: (TIM) Wows. These are waves that contain a nonzero E;.. ' but H, = 0. 3. Tm~nuerse Electric (TEE) W a m . These are waves that contain a nonzero ?I= .

hut IS, = 0.

The propagation charac:eristics of the vorious types of waves are different; they will be discussed in,subsequc;nt subsections.

10-2.1 Transverse ~lectrotna~net ic Waves

Since E, = 0 and Hz = 0 for TEM waves within a guidc, we see that Eqs. (10-9) through (10-12) constitute a set of trivial sol~t ions (all field components van~sh) unless the denominator h' also equals zero. In other words, TEM waves exist oniy when

+ k' = 0 (10-14) or

;lTEM = jk =ju JF, (10-15)

which is exactly the same expression for the propagation constant of a uniform pime wave in an unbounded inedium characterized by constitutive parameters s and 11.

We recall that Eq. (10-i5) also holds for a TEM wave on a lossless :ransmission line. It follows that the \.elocity of propagation (phase velocity) for TEM waves is

We can obtain the ratio between E:! and H; from Eqs. (10-7b) and (10-8a) by setting E, and H , to zero. This ratio is called the wcive impedance. We have

r !

.. , . . 448 . WAVEGUIDES AND CAVITY RESONATORS 1 10

which becomes, in view of Eq. (10-15), .

, . > (10-18)

We note that Z,, is the same as the intrinsic impedance of the'dielectric medium, as given in Eq. (8-25). Equations (10-16) and (10-18) assert that the phase uelocity and the wave impedance for TEA4 waves ore iirdependent of rl~e frequency of the ivnves.

Letting EP = 0 in Eq. (10-7a) and H p = 0 in Eq. (10-8b), we obtain

Equations (10-17) and (10-19) can be combined to obtain the following formula for a TEM wave propagating in the + 2 direction:

which, again, reminds us of a similar relation for a uniform plane wave in an unbounded mcdium -see Eq. (8 -24).

Single-conductor waveguides cannot support TEM waves. In Section 6-2 we pointed out that magnetic flux lines always close upon themselves. Hence. if a TEM wave were to exist in a waveguide, the field lines of B and H would form closed loops in a transverse plane. However, the generalized Ampere's circuital law, Eq. (7-38b), requires that the line integral of the magnetic'field (the magnetomotive force) around any closed loop in a transverse plane must equal the sum of the longitudinal conduction and displacement currents through the loop. Without an inner conductor, there is no longitudinal conduction current inside the waveguide. By definition, a TEM wave does not have an E, component; consequently, there is no longitudinal displacement current. The total absence of a longitudinal current inside a waveguide leads to the conclusion that there can be no closed loops of magnetic field lines in any transverse plane. Therefore,. we conclude that TEM waves cannot exist in a single-condttcror hollow (or die1ectric;filled) waceguide of any shape. On the other hand, assuining pei$ect conductors, a coaxial transmission line having an inner conductor can support TEM waves: so can a two-conductor stripline and a two-wire

,transmission linc. Wllcn (17c conductors havc losscs, waves along transmission iirics arc strictly n o lo~igcr 'I'LM. 11o1cd i l l S cc t i o~~ 9 2.

10-2.2 Transverse Magnetic Waves

Transverse magnetic (TM) waves do not have a component of the magnetic field in the direction of propagation. H, = 0. The behavior of TM waves can be analyzed by solving Eq. ( I 0 . 5 ) l i ) ~ E, subject to tlic hountl:lry conditions o r tlic guiclc :lnd using

:diuni, , i-locity waves. I

. .

ula for

- 2 we . T E M ! loops -38b),

:round mduc- :. there . TEM .a1 dis- eguide lnes in ,r pl a

2 other :r con- :o-wire In lines

P

field in :zcd by ii using

Eqs. (10-9) ihrough (10-12) to determine the other components. Writing Eq. (10-5) for E;, we have L

ViyE: + (y2 + k?)~: = 0 ,.; . (10-21)

- or

Equation (10-22),is a second-order partial diffdrential equation, which can be solved for E:. In thitsection we wish only to discuss,the general properties of the various wave types. he actual solution of Eq. (10-22) will wait until subsequent sections when we examine patticu!ar wavcguidcs.

For TM waves, we set Hz = 0 m Eqs. (10-9) through (10-12) to obtain

It is convenient to combi1.e Eqs. (10-23c) and (10-23d) and write

L

where

denotes the grkdient of 1:: in the tmnsvcrse plane. Equation (10-24) is a concise formula for finding E: an.i E,O from Ep.

-. The transverse components of magnetic field intensity, H, and H,, can be deter-

mined simply from Ex a rd E, on the introduction of the wave impedance for the TM mode. We have, from Eqs. (10-23),

--.-

It is important to note thai Z.,. is not cquil? to jw,u/y, becausc y for TM waves, unlike YTEM, is not equal to jo ,/a. The following relation between the electric and magnetic

field intensities holds for TM waves: 1

L J

Equation (10-27) is seen to be of the same form as Eq. (10-20) for TEM waves. When we undertake to solve the two-dimensional homogeneous Helmholtz

equation, Eq. (10-22), subject to the boundary conditions of a given waveguide, we will discover that solutions are possible only for discrete ralues of h. There may be an infinity of these discretc valucs. hut solutions ilre not possible for all values of 11. The values of11 for which a solution of Eq. (10-22) exists are called the c h i f r i r c t ~ r i ~ f i ~ uolllrr or eioe~~crilues of the boundary-value problem. Each ol the rigenvalues determines the characteristic properties of a particular TM mode of the glven waveguide.

In thc following sections we will also discover that the ei~envalues of the various wilveguidc problcms are real numhcrs. From Eq. (10-1:) we have

7 = ,/I,: - 1;: --- r? 7

--- = , jh- - (?)-/A€. (10-3)

Two distinct ranges of the values for the propagation constant are noted, the dividing point being y = 0, where

w:,u E = 112 (10-29)

The frequency,.j,, a t which 7 = O is called a cu to l j ' j r cq~~~~ncy . Ti" uuhe ($,I; for a partieuior mode in a wuueyuidr dcpeads or1 lie eige~lcalur uj this mode. Using Eq. (10-30), we can write Eq. (10-28) as

(10-31)

The two distinct ranges of p can be defined in terms of the ratio (f/jd2 as compared to unity.

) ( > 1 , > . i n this rrngr, wzy r ; /I' and 7 is imaginary. w e have. from

Eq. (10-2% . - - - -- . -

= jp = j k dl - ( ! ) I< =,k Jm. (10-32)

It is a propagating mode with a phase consrant 0: I

:s. llholtz :de. we nay be 2; oi 11.

terisfic , detcr-

't cguide. - - various

(10-28)

d i v i e g

(16 ,)

( I 0 - 30)

J; /or u sing Eq.

(10-31)

-0111 p;11.ctl

: ax , from

, r 3 2 )

(ld-33)

-*\b. % .srr.: r-?*.;r....a,s~.'.*L..*. *%$ * .+ I +\ I , L+,, 1- *.sf P a h * * .*P*. j l ~ l l ? l u ~ ~ ~ 4 W ? - ~ .*llTr - j*' b W F C ' . *m'''qL'*wwpr

1::. +, 10-2 GENERAL WAVE BEHAVIORS ALONG UNIFORM QU~D~NG STRUCTURES 451 4,:: 'i'. , ;

The corresponding wavelength in the guide ifi

where 2 n 1 ' u

i a m c m pl- (10-35) "+*....k,.* ,/z* a f 2

is the wavelength of a plane wave with a frequency f in an unbounded dielectric medium characterized by p and s, and LL = I/& is the velocity of light in the mcdiurn.

The phase velocity of the propagating wave in the guide is

We sce from Eq. ( 1036) thilt thc phase vclocity within a wavcguidc is always higher thzn that in a.1 unbounded mcdiurn and is frequency-dependent. Hence single-conductor wuucguides ure dispersive rrut~smissiot~ systems. The group velocity for a propagating waie in a waveguide can be determined by using Eq. (8-j2):

Substitution of Eq. (10-32) in Eq. (10-26) yields

L I

The wuae irqwduizce 1 f'p'apuyotirig TAM tnoddrs i~ u wur.eg.de is picrely resistive am! is uiwuys less tho.i the ir?rrinsi i i~ lp iu t l ee of the dieiecri-ic rnedietn. The vnri- ation of Zm versa j : / ; for f > /, is sketched in Fig. 10-2.

Fig. 19-2 Normalized wave irnped- ances for propagating TM and TE waves.

b) ($1' < 1, or f < /,. When the operating frequency is lower than the cutoff

frequency, y is real and Eq. (10-31) can be written as

which is, in fact, an attenuation constant. Since all field components contain the propagation factor e-?' = e-", the wave diminishes rapidly with r and is said to be evanescent. Therefore a waveguide exhibits the property of a high-pass jlter. For a given mode, only waves with a frequency higher than the cutoff

*\, frequency of the mode can propagate in the guide. Substitution of Eq. (10-39) in Eq. (10-26) gives the wave impedance of

TM modes for f < f;.:

-1

Thus, the wave impedance of evanescent TM modes s t freq>encies below cutoK is purely reactive. indicating that tlicrc is no powcr l low associ:~tcd wit11 cv;inusccnt wa\'cs.

10-2.3 Transverse Electric Waves

Transverse electric (TE) waves do not have a component of the electric field in the direction of propagation, Ez = 0. The behavior of TE waves can be analyzed by first solving Eq. (10-6) for H,:

Proper boundary conditions at the guide walls must be satisfied. The transverse field components can then be found by substituting H z into the reduced Eqs. (10-9) through (10-12) with 4, set to zero. We have

SHP tr; = -L- h 2 C),

. . . , ;.. . ,.. !i 1!$i'

7e' cutoff ' : ., ,, ,

(10-39)

ntuin the id is said high-pass :he cutoff

' >. :dance of

(10-40)

3 W c p f f ;meL ~ l t

-Id in tlic Jyzed by

(10-41)

xansverse . p . (10-9)

( 10-42a)

( l 0 P b )

( 10 -*k)

( 10 -43d)

+ : -. . 10-2 1 GENERAC W A J ~ BEHAVIORS ALONG UW~FORM GUIDING STRUCTURES 453 .i'

. c. _ ).- '. . ,

I

Combining Eqs. (10-42a) and (10-42b). we obtain

We note that Eq. (10-43) is entirely similar to Eq. (10-24) for TM modes. The transverse components of eiectric field intensity, EO and E;, are related to

those of magnetic field intensity through the wave impedance. We have, from Eqs. (10-42% b, c, and d),

Note that Z,, in Eq. (10-44) is quite different from Z,, In Eq. (10-36) bemuse ;) for TE waves. unlike ynln is ,lor cq l~a l to ,or fi. Equations ( 1 0 4 2 ~ ) . (10 -42di. ad ( I 0 44) c;in t w w bc c o ~ ~ l l ) i ~ l ~ d IO give L I I C SOIIOWIIIC VCCLOI lil.~:iuIa:

Inasmuch as we hahc not changed the relation between 7 and 11. Eqs. (10-28) through (10-31) pertaining to TM waves also apply to TE waves. There are also two distinct ranges of 7, depending on whethdr the operating frequency is higher or lower than thc cutort 'frcqu~nc~, J C , givcn in Eq. (10-30).

a) ( > 1, or j' . In this range 7 is imaginary. and we have a propagaring

mode. The expression for 7 is the same as that given in Eq. (10-32):

(1 0-46)

Consequently, the formulas for 1, )L,, u p , and u, in Eqs. (10-33). (10-34). (10-36). and (10-37), respectively. also hold for TE waves. Using Eq. (10-46) in Eq. (10-44), we obtain

I .

I b) ($)2 < 1, or / <I:. In this case, 7 is real and we have an evanescent or non-

propaghting mode

. ,

Substitution of Eq. (10-48) in Eq. (10-47) gives the wave impedance of TE modes for f <f,.

which is purely reactive, indicating again that there is no power flow for evanescent waves at f < h.

Example 10-1 (a) Determine the wave impedance and guide wavelength at a frequency equal to twice the cutoff frequency in a waveguide for TEM. TAM, and TE modes. (b) Repeat part (a) for a frequency equal to one-linlf-olrhe cutoff frequency.

a) At f = 2A. which is above the cutoff frequency, we have propagating modes. The appropriate formuias are listed in Table 10-1.

Table 10-1 Wave Impedances and Guide Wavelengths for j' > J':

1 TEM j/ n = ~ ~ E

fre- i TE

v 0 " Fig. 10-3 An w-b diagram for waveguide.

b) At f = J,/2 < f,, the waveguide modes are evanescent and guide wavelength - has no significance. We now have

We note that Z,,, does not change with frequency'bewuse TEM waves do not exhibit a cutoff propeity. Both Z,, and Z& become imagnary for evanescent modes at f c /;; their values depend on the eigenvaluc 11, whlch is a characteristic of the particular TM or TE mode.

For propagating modes, 7 = j/i and the variation of /j' versus frcquency determines the characteristics of a wave along a guide. It is therefore useful to plot 2nd examine an W-fl diagram.' Figure 10-3 is such a diagram in which the dashed line through the origin r ep rc sds the w-8 relationship for TEM mods. The constant slbpe of this stmight iine is iu/b = u = I,'&, which is the same as the velonty of light in an unbounded dielectric medium with cbnstitutive parameters p and 6.

The solid curve abovr the dashed line depicts a typical w-P relation for either :I TM o r ; I Ti7 ~ ~ l ' O ~ ~ : l g i l ~ i l l r ~ I I I O ( ~ C , givc11 l )y J{CJ, (10 33). We GIII writc

NI I

The a-,/3 curve intersects ihe W-axls ([i = 0) at o = m,. The slope of the line jo~nlng the origin and any point. L U C ~ as P. on the curve is equal to the phase velocity, up, for a particular mode having 3 cutoff frequency /; and operating at a particular

' Also referred to as a Biillouin drayram.

Fig. 10-4 Relation between attenuation constant and operating frequency for evanescent modes (Example 10-2).

frequency. The local slope of the w-/3 curve at P is the group velocity, u,. We note that, for propagating TM and TPwaves in a waveguide. up z u and ug < u. In fact, Eqs. (10-36) and (10-37) show that

As the operating frequency increases much above the cutoff frequency, both 11, and ug approach u asymptotically. The exact value of o. depends on the eigenvalue h in Eq. (10-30)- that is. on the particular TM or TE mode in a waveguide of a given cross section. Methods for determining h will be discussed when we examine different types of waveguides.

Example 10-2 Obtain a graph sliowin$ thc relation between the attenuation constant n and the operating frequency f for evanescent modes.

Solution: For evanescent T M or TE modes, f < and Eq. (10-39) or (10-48) applies. We have

Hcncc thc graph of(j.r/h) plotted versus/ is a circle ccntcicd at the origin and having a radiusl,. This is shown in Fig. 10-4. The value of n for any/' < j; can be found from this quarter of a circle.

10-3 PARALLEL-PLATE WAVEGUIDE

In Section 9-2 we discussed the characteristics of TEM waves propagating along a parallel-plate transmission line. It was then pointed out. and again emphasized in subsection 10-2.1, that the field behavior for TEM modes bears a very close resemblance to that for uniform plane waves in an unbounded dielectric medium. However, TEM modes are not the only type of waves that can propagate along

Je note ?.

In fact, I

, and u,

Lie /r

In con-

having :d from

r'

I a ,IZU,. ~n IOSC re- irdium. : along

perfectly conduc& p~mllel-plates separated by a dielectric. A parallel-plate wave- '" ' 'guide can also support' TM and TE waves. ?he characteristics of these waves are

examined separately in following subsections. +

10-3.1 TM Waves between parallel Plates I '

Consider the paallel-plate waveguide of two perfectly conducting plates separated by a dielectric medium with constitutive parameters 6 and p, as shown in Fig. 10-5. The plates arc assbmed to bc infinitc in extent in thc x-direction. This is tantamount

' to assuming that the Aelds do not vary in the x-direction and that edge effects are negligible. Let us suppose that TM waves (Hz = 0) propagate in the + z direction. For harmonic time dependence, it is expedient to work with equations relating field quantities with the common factor ej(''"-yz) omitted. We write the phasor Ez(y, 2) as . ~:(y)e-~ ' . Equation (10-22) then becomes

d 2 ~ ; ( y ) + h2E;(y) = 0. cly2 (10-53)

The solution of Eq. (1b-5:3) must satisfy the boundary conditions

E:'(y)=O a t y = O and y = h .

From Section 4-5 we cor-clude that E:(y) musr be of the Sollowing form (h = n q b ) :

where the a m p h d e A,, depends on the strength of excitation of the particular TM wave. The only other nonzero field components are obtained from Eqs. (10-23a) and (10-23d). Keeping in mind that ZE,/?x = 0 and omitting the e - p factor, we have

jar (n:) H ~ ( Y ) =- A, cos - h

'Fig. 10-5 . waveguide.

Ad infinite parallel-plate

, , .;. .

----- . , #,:" . a , . . . . - . . -

. . . . . . . . .

. , - . . . \ . . . . < . -. . . - : , . . , .; ,:: :. :. . , . , ' " ' .. . . . . "., ... -,..+. . ...:. ::'.. , . . ' . . ' -'-: . . . _ . . .

- . . . . . . , . . . - , - -

*.I&~.*l?*.;< .. . .. -.. , V:,J;* - : ,, .. ,: ,;-. . , . , . . . . , :, .: , ,,.,,&$ ,...a ;::h..; &... ;,.'.-.,.y: .$., +%>! .,,.... A,"",.! :i: .,.. , , . -

.- : , ' . .\ . . 6 8 . : WAVEGUIDES AND C A V l n R E S ~ ~ ~ ~ ~ ~ & 1 10 , .

2'. ,

The y in Eq. (10-54c) is the propagation constant that can be determined from Eq. (10-28):

Cutoff frequency is the frequency that makes y = 0. We have

which, of course, checks with Eq. (10-30). Waves with f > propagate with :I phase constant A given in Eq. (10-33): and waves with / 5 J; are evanescent.

Depending on the value of n. here are different possible propa,. o ~ t i n g TM modes (e~genmodes) corresponding to l l~c iiilhrent cigcll~;~lucs /I. Tllar. ~llcrc ;lrc.lllc Tkl , mode ( 1 1 = I) with cutolYfrrquency ( j ; ) , = l/ll~,/~a. the TMl.mode (11 = 2) with =

iC

I/hdli+ and so on. Each mode has its own characteristic phase constant, guide wavelength, phase velocity. group velocity, and wave impedance; they can be determined from. respectively, Eqs. (10-33), (10-34). (10-36), (10-3i), and (10-38). When n = 0, E, = 0, and only the transverse components H , and E,, exist. Hence TIM, mode is the TEM mode, for which j, = 0. The mode having the lowest cutoff frequency is called the dornindnt mode of the waveguide. For parallel-phte waveguides, the domi- nanr rnorie i s rlw TEM inode.

Example 10-3 (a) Write the instantaneous field expressions for TM, mode in a parallel-plate waveguide. (b) Sketch the electric and magnetic field lines in the yz-plane.

Solution

a) The instantaneous field expressions for the TM, mode are obtained by multiplying the phasor expressions in Eqs. (10-54a), (10-54b), and (10-54c) with eJ("'-a' and taking the real part of the product. We have, for 11 = 1,

where

ncd Ii-onl

i ;. (10-55)

--(10-56)

h a phasw-

.M modes the TM,

1th ( I ; ) , = :nt, gutdc be @r-

cn M, jde

qu$ncy is the domi-

node in a I pz-plane.

5y multl- -54~) with

(10-57a)

(10-57b) r

I ! 7 4

(10-58)

. " , . . b) In the y-z E has both a y and a z component, the equation of the electric

, :i-l .,$eId lines w a given t can be found from the relation: . ,- % . I.*

, - , * , . .: dy ddz - = -. (10-59)

E, EL

For example, at t = 0, Eq. (10-59) can be Written as

d ~ ' E,(Y, I; 0) - - = - -@ cot (y) tan f i z , dz E,(y,z;O) le

which gives the dope of the electric field lincs. Equation (10-60) can be integrated to give

which is the equhtion of the electric field line for a particular yo at ; = 0. DiLrent values of y,, givc dilfcrcnt loci. Severill i l ch electric licld lincs ;ire d r ~ w n in Fig. 10--6. The ficld lines rcpwt tl~cmaclvcs iur every cliangc of4r by ?n rad.

Since H ha3 only an s component, tNc magnetic field lincs are everywhere perpendicular to the y-; pime. For thc TM, mode at t = 0, Eq. (10-57c) becomes

The density of h, lines varier as cos in i /b) in the y direction and as sin 0: in the z direction. This is also skctched in Fig. 10-6. At the conducting plates (y = 0 and y = h). there arr. surface currcnu because of a discontinuity in the tangential magnetic field and : urface charges because of the presence of a normal e1ec:ric field. (Problem 10-1).

- Electric t':c!d lines. a 8 Mngnctic field lincs (.\--asis inld thc paper).

Fig. 10-6 Field llnes I J r TM, mode in parallel-plate waveguide.

. . . / . . /

- 460 . WAVEGUIDES AND CAVITY RESONATORS / 10 . . , . .& .- r .

. ,. !

9 Y Example 10-4 Show that the field solution of 9 propagating TM, wave in a parallel-

.!, , , , ,piate waveguide can be interpreted as the superposition of two plane waves bouncing back and forth obliquely between the two conducting plates.

Solution: This can be seen readily by writing the phasor expression of EP(y) from Eq. (10-54a) for n = 1 and with the factor e-jB' restored. We have

A1 = - [ e - j ( B : - n ~ l b ) - - j ( P : + n ~ l h t I. (10-G3) 2j 4 '.

From Chapter 8 we recognize that the first term on the right side of Eq. (10-63) represents a plane wave propagating obliquely in the + z and -y directions with phase constants ,!3 and n/b respectively. Similarly, the second term represents, a plane wave propapting obliquely in the +: and + y directions with the same phase constants and d h as those of the first plane wave. Thus, a propagating TM, wave in a parallel-plate waveguide can be regarded as the superposition of two planc waves. as dcpictcd in Fig. 10-7.

In Subsection 8-6.2 on reflection of a ~arallelly polarized plane wave incident obliquely at a conducting boundary plane, we obtained an expression for the longitudinal component of the total E l field that is the sum of the longitudinal components of the incidcnt Ei and the retlectcd E,. To adapt the coordinate designations of Fig. 8-10 to those of Fig. 10-5, ..c and z must be changed to z and - y respectively. Wc rcwritc E, of Eq. (8--86a) as

Comparing the exponents of the tcrms in this equation with those in Eq. (10-63), we obtain two equations:

/3, sin Bi = p (10-64a) X . p, cos Oi = -- b

Fig. 10-7 Propagating wave in parallel-plate waveguide as superposition of two plane waves.

t.10-63)

i 10-63) ns with "- a plane se con- aave in

M'3 VCS,

n

i C l r .L

he .- : I corn- xations ;~ively.

p =

which is the same as Eq. (10-58), and

n /Z COS 0, = - -

P,b - 26' (10-65)

where I = 2n//3, is the wavelength in the unbounded dielectric medium. We observe thLt B solution o l Eq. (10-65) for Oi exists only whcn L/Zb i I. At

/./Zb = I . or /' = I I / ~ = l / ? b , / ~ . whicll is the cutoff frcqucncy in Eq. (10-56) for 1l = 1, cos Oi = I. and Oi = 0. This corresponds to the case when the wves bounce .

back and forth in the j9 direction, normal to thc prailei plates. ilnd there is no propagation in the : direction ( b = 11, sin Oi = 0). Propagation of TM, mode is possibie only when /. c i, = 2b or f > ji.. Both cos Oi and sin Oi can be expressed in terms of cutoff frequency /:.. FPorn Eqs. 11 0 - 65) and ( 1 0-64a) wc htvc

and

sin Qi =5=2=/%. /.g I$ (10-66b)

Equation (LO-66bj is in agreement with Eqs. (10-34) and (10-36).

10-3.2 TE Waves between Parallel Plates I For transverse eiectnc waves. Ez = 0, we solve the following equation for HP()),

which is a simplified versim of Eq. (10-41) with no *-dependence.

We note that H,(R :) = H :( y)e-". The boundary conditions to be satisfied by H:[)I) are obtained from Eq. (10-42c). Since Ex must vanish at the surfaces of the conducting plates, we require

H:( g) = B,, fos (F) ,

a . i <. c . , . .. . / . . I . , '. .

L ' . 462 WAVEGUIDES AND CAVITY RESONATORS I 10

' 5 .

. , where the amplitude B,, depends,on the strength of excitation of the particular TE . . * % I _ wave. We ~ b t a i n the only other nonzero fidd components from Eqs. (10-42b) and

(10-42c), keeping in mind that dHz/2x = 0:

The propagation constant y in.Eq. (10-68b) is the same a3 that for TM waves given

3 '. in Eq. (10-55). Inasmuch as cutoff frequency is the frequency that makes y = 0, tlre c u t o ] ' j i c q u a ~ j ~ for rlic TE, rlrotlc iir ( 1 patrllcl-pltrrc wreguidc is estrcrl!i the strrnr as t l lu t jbr the TM, mode gicen irt E y . (10-56). For n = 0, both H , and E, vanish; hence the TE, mode does not exist'ln a parallel-plate waveguide.

Example 10-5 (a1 Write the instantaneous field expressions for the TE, mbde in a parallel-plate waveguide. (b) Skctch thc elcctric m d m a ~ ~ i c t i c field lincs in the J*-:

plane. .-. Solution

a) The instantaneous field expressions for ;he TE, mode are obtained by taking the real part of the products of the phasor expressions in Eqs. (10-68a), (10-68b), and (10-6Sc) with ej(""-4''. We have. for n = 1,

H,(y , z : t ) = B1 cos - cos (at - Pz) (lid) (10-69a)

where the phase constant P is given by Eq. (10-58), same as that for the TM, mode.

b) In the y-z plane E has only an x component. At t = 0, Eq. (10-69c) becomes

' Thus the density of Ex lines varies as sin (xylh) in the y direction and as sin /?i in the i direction; Ex lines are sketchcd as dots and crosses in Fig. 10-8.

The magnetic field has both a y and a z component. The equation of the magnetic field lines at t = 0 can be round from the following relation:

d y I! ( v 2 : 0) /lh - = - tan - tan llz. dz H z ( y , z ; 0) n (7)

. 0-~SC)

:s given = 0, rhr 1e same vanish; ,,

.,

~ d e in a the y-z

P %In, e 0-boo),

I 0-69a)

10-69 b)

10-69~)

2e TM,

.ornes

i 10 -70)

1 >In /k f-

I L ' ::

\ 10-71)

---- Magnetic flcld lines. O 8 ~ lcc t r i c fieid lines (x-axis into the paper).

Fig. 10-8 Field line; for TE, mode in parallel-plate waveguide.

Upon integration, Eq. (10-71) gives

sin (nb 11)) = ,-2%

SIII (ny/h) '

which is the equation of thc magnetic field line for a particular y o at z = 0. Several such lifles are drawn in Fig. 10-8 for different values of yo. The field lines repeat themselves for every change of f lz by 2x.rad.

10-3.3 Attenuation in Parallel-Plate Waveguides

Attenuation in any wa~c.guide.(not just the parallel-plate waveguide) arises i'rorn two sources: lossy dielectric and imperfectly conducting walls. Losses modify the electric and magnetic fields within the guide, making exact solutions difficult to obtain. However, in practical wa1:eguides the losses are usually very small, and we will assume that the transverse field patterns of the propagating modes are not affected by them. A real of thc propagation constant now appears as the attenuation constant, which accounts for power losses. The attenuation constant consists of two parts :

.,u = ud + z,, (10-73)

TEM Modes The attenuat.on constant for-TEM modes on a parallel-plate transmission line has been discwed in Subsection 9-33. From Eq. (9-72) and Table 9-1

we have approximately , < . . , , I where E, ,u, and.a are, respectively, the permittivity, permeability, and conductivity of the dielectric medium. In Eq. (10-73a) if = is the intrinsic impedance of the dielectric if the dielectric is lossless. Also from Eq. (9-72) and Table 9-1 we have

where a, is the conductivity of the metal plates. We note that. for TEM modes, r, is independent of frequency. and cc, is proportional io J?. We note further that a,, -. 0 as a + 0 and that r, -+ 0 as a, 4 cc. as expected.

TM Modes The attenuation constant due to losses in the dielectric at frequencies above ji can be found from Eq. (10-55) by substituting E,, =-e.+(aijw) for 6. We have

Only the first two terns in the binomial expansion for the second line in Eq. (10-74) are retained in the third line under the assumption that

From Eq. (10-56) wc scc t h a t

With this relation, Eq. (10-74) becomes


-- -- .... -ES- . . . - ? I)- - -" ..... :

. ..*:-:,*:<.._ . (* < L a ".',: 6 . - .. 8 -.: .., .&fmJq' . . . . ' . 1, .. .....,... :r ... ~rr*'r:*y~uthi*--u\~*-** Lr'su**'ii.' .A'‘..'..

10-3 I%IRALLEL-PLATE .wAvEciwx 465 . . . ' 3 . >-.+*;:.+ ': : - I..

s = w& J- iradlm). (10-76) - . I a

Thus ad for TM mod& decreases when keqienky increases. .. t r

To find the atterluation constant due ta losses in the imperfectly conducting plates, we use Eq. (9-70), which was derived from the law of conservation of energy. Thus,

. (10-73a) ; 1 , , -* :<'I: ,ndudtivity - - . ' !:"";' '

mce of the we have

.- (10-73b)

modes, a, xther th8i

,

'requencies e. Wc have

fi

(10-74)

- kq. (10-74)

J1

(10-75)

P k ) S(, = -

2 ~ ( $ (10-77)

where P(r) is the time-average power flowing through a cross section (say, of wldth w) of the waveguide, Bnd P,(3 is the time-average power lost in the two plates per unit length. For TM modes we use Eqs. (10-54b) and (10-54c):

. I'hc surfxc curicnt dwsil:cs on illv aplicl- ;md lower plates hilve thc silnie n,agniiudc. On the lower plate where y = 0, wi: havc

The total power loss per unit length in two plates of width bv is

Substitution of Eqs. (10-Xa) and (10-78b) in Eq. (10-77) yields

where, from Eq. (9-26b),

The use of Eq. (10-801 in Eq. (10-79) gives the explicit dependence of r , on j for TM 11locit.s7- ._ -

A sketch of the no&alize(lz, is shown in Eig. 10-9, which reveals the existence of a minimum.

i attcn- finite :es in * ...

I 10-4 RECTANGULAR ~ A ~ E G U I D E S

- The analysis of pdnhel-plate waveguides in Section 10-3 assumed the plates to be of an infinite extent it the transverse x direction; that is, the fields do not vary with x. In practice, these plates are always finite in width, with fringing fields at the edges. Electromagnetic endgy will leak through the sides of the guide and create undesirable stray couplings to other circuits and systems. Thus, practical waveguides are usually uniform structures of a cross section of the enclosed variety. The simplest of such cross sections, in ternis of ease both in analysis and in manufacture, are rectanguhr and circular. In this section we will analyze the wave behavior in hollow rectangular waveguides. Circular waveguides will not be treated in this book, because to do so requires a knowledge of the properties of Bessel functions. Renders possessing such

- knowledge. however; would have little difficulty following an .~n;iiysis of circular . s e v q u i d r . ~ in more advanced books. bcc;luse the procedure is the same as d e m ~ b e d here. Rectangular waveguides are much more commonly used in practlce than circular waveguides.

In the following diw.r:i~iotl. wc .lr:iw 011 ~i ie n ~ i i ~ c r ~ i ~ l 111 S C C I I , ~ ~ 10 1 con'crnlng ~ C I ~ C I X ~ wavc bcliaviorr along unrllml guiding struclurcs. l'rupapation of nrnc- harmonlc waves in the + z direction wirh a propagation constant 7 is considered. -.. I M and TE modcs will be discussed scpamtcly. As wc h;~vc notcd pmv~ously, TE\I

wave, cannot exist B a single-conductor hollow or dielectric-tilled wnvegulde.

10-4.1 TM Waves in Rectangular Waveguides

Consider the waveguide ketched in Fig. 10-10, with its rectangular cross sec:ion of sides u and b. The enclosid dielectric medium is assumed to have constitutive parameters r and p. For TM waves, H , = 0 and Ez is to be solved from Eq. (10-22). Writing E&, Y9 4 as

. . _ . I , , . I.. 7. , " , : . ' .:. . _-. . . .. .. , ,'. ?. .y . .

. . . ' .. , .. I ,. . . . . . . . . , .

I . . . a

, . . . . i . > : , ' . . : - . ' ., .-.;,. ,a ?iT .-... ' l . , ' . . I;: . . ; .. , . .;-

, , . . . 468 " WAVEGUIDES AND CAVITY RESONATORS 1 10 . . , :-c. - . . - . ," . . ,. -

. . we solve the following second-order partial differential equation: %

C

, J

a2 aZ ( + + h Z (10-85)

it f 4

Here we use the method of separation of variables discussed in Section 4-5 by I ?C

letting s'

Substituting Eq. (10-86) in Eq. (10-85) and dividing the resulting equation by X ( x ) Y ( y), we have

I t Now we argue that, since the left side of Eq. (10-87) is a function of x only and the 1

I right side is a function of y only. b6th sides must equal a constant in order for the equation to hold for all values of .K and y. Calling this constant k:, we obtain two 4

separate ordinary diffcrcntinl equations:

2 Y ' ) - 7 - ddF + ~ ; Y O . ) = O . (10-89)

- where k2 = p - ,p

'c' (10-90) The possible solutions of Eqs. (10-88) and (10-89) are listed in Table 4-1.

Section 4-5. The appropriate forms to be chosen must satisfy the following boundary F , conditions.

1. In the .K direction: EP(0, y) = 0

. EP(a, y) = 0. 2. In the y direction:

E~(.Y. 0) = 0

Ey(s , 6) = 0.

Obviously, then, we must choose:

X ( x ) 111 the form of sin li,.~,

Y ( y ) in the form of sin k,y,

7 . ,. -d . 466' WAVEGUIDES AND CAVITY RESONATORS / 10 f

I I TE modes

Fig. 10-9 NormaIized attenuation constant due to finite conductiJity of the plates in parallel-plate waveguide.

TE Modes In Subsection 10-3.2 we noted that the expression for the propa_eation constant for TE waves between parallel plates is the same as that for TM waves. It follows that the formula for a, in Eq. (10-75) holds for TE mode~ i iwe l l .

In order to determine the attenuation constant a, due to losses in the imperfectly conducting plates, we again apply Eq. (10-77). Of course, the field expressions in Eqs. (10-68a). (10-68bh and (10-68c) for TE modcs must now bc used. We have

A normalized a, curve based on Eq. (10-53) is also sketched x, for TM modes, a, for TE modes does not have a minimum tonically as f increases.

in Fig. 10-9. Unlike but decreases mono-

10-4.1 Rectan

and the proper so~utibn for E;(X, y ) is

E:(x, y)' = E, sin (10-92)

\ and the .

: for the :sin two

I . ,

From Eq. (10-go), we have

- The other field components arc obtained from Eqs. (10-23a) through (10-23d):

where

, y mn

E , o ( ~ , yi = -2 (7) E , cos x ) sin (y )) . r

y nx ~ e ( x . 1.: = -p (,) E, sin

x) cos (7 y ) . ,

joc nn H ~ L Y , y = ( T ) ~ , , sin (yx) cos (7 y)

Every combination of the integers ni and 11 defines a possibic mode that may be designated as the TM,,, nlode; thus there are a double infinite number of TM n o d s . The first subscript denotes the number of half-cycle variations of the fields in the x-direction, and the second subscript denotes the number of half-cyc!e variations of the fields in they direction. The cutoff of a particular mode is the condition that makes y vanish. For the TM,,,,, mode, the cutoR frequency is

L

which checks with Eq. (10-30). Alternatively, we may writo ----

where i, is the curoj"~uve1enyth.

, For TM modes, neither m nor n can be zero. (Do you know why?) Hence, TM,, , I , . , . . mode has the lowest cutoff frequency of all TM modes in a rectangular waveguide.

, . ' The expressions for the phase constant j3 and the wave impedance Z, for propagating modes in Eqs. (10-33) and (10-38), respectively, apply here directly.

Example 10-6 (a) Write the instantaneous field expressions for the TM,, mode in a rectangular waveguide of sides a and 6. (b) Sketch the electric and magnetic field lines in a typical x-y plane and in a typical y-z plane.

a) The instantaneous field expressions for the TM,, mode are obtained by multiplying the phasor expressions in.Eqs. (10-92) and (10-94a) through (10-94d) with e ~ ( u t - P : ) and then taking the real part ofthe product. We have, for in = n = 1.

----- E,(x, y. 3; t) = $-(;)TO sin (: x).-cos (% y ) sin (wt - pi) (10-97b)

. .

E-(x, y, z; t) = E, sin - .u sin - y cos (o t - pz) (r ) (; (lo-97c)

HJx, y, z; t) = -- ma (?) E,, sin (: X) cos (F g ) sin (or - Bz) (10-97d) h' b

H,(.x, y, z ; t) = T E , cos - x sin - y sin ( o t - jz) (10-97e) 11 - ' ( ( ) (; )

where

b) In a typica I x-y plane, the slopes of the electric field and magnetic field lines are

@IE = 9 tan (: x) cot (; .y)

h (;IH = -; cot ( z x) tan (i y) - These equations are quite similar to Eq. (10-60) and can be used to sketch the E and H lines shown in Fig. 10-ll(a). Note that from Eqs. (10-99a) and (10-99b)

:c field

'\ multi- - - ) 94cl)

17 = 1,

.ines are

* I \ ' ; .::. ? -. 3

.. -- ,: . i r' ' ; J o ~ /. RECTANGULAR WAVEGUIDES 471 ... 1

- Electric field l i ,m -- - - - Magnetic field lines

Fig. 10-11 Field lines for TM,, mode in rectangular waveguide.

. i ~ n d u t i n g that E and H lines are everywhere perpendicular to one mother. Note also that E liiles arc normal and that H lines are pir:illel to conducting guldc walls.

Similarly, in a i j p i ad y r p h c , say. Br .Y = u/2 or sin (n.i/(r) = 1 sod cos ( 7 r s / ( 1 ) = 0. we h;.vc

(?)[: = $ (i) Cot (i y) tan cwt - pz i ,

and t i has only an s-component. Some typical'^ and H lines nrs drawn in Fig. 10-1 l(b) for r = 0.

10-4.2 TE Waves in Rectangular Waveguides

For transverse electric waves, .!iz = 0, we solve Eq. (10-11) for I f z . We write

H,(x, y, z) = H;(X, ~ l ) e - ~ " , (10-100)

where H:(x, 2 ) satisfies the following second-order partial differential equation:

Equation (10-101) is seen to be of exactly the same form as Eq. (10-85). The solutron for R;(I. fl must s m f y :he fcllcwing !xmdar i cuiidit~onb. .. - -.

1. I11 the s-dircction:

2. In the y direction: ,.-.

-= aH;o O & O ~ i t y = o ZY (10-102c)

aHp -=

JY 0 (Ex = 0) at y = h. (10-102d)

It is readily verified that the appropriate soiution for H-?(~ , )-) is

(10-103)

4 .

The relation between the eigenvalue hand (mn/a) and (nnib) is the same as that given in Eq. (10-93) for TM modes.

The other field components are obti~ined from Eqs. (10-42a) tlirough (10-42d):

j o p nn E'(-y. Y ) = - (.) H o COS (~p ) sin (y ,)-\ (10- 1 h'

. ,

E,:(-Y,J) = -- "o" (I:) - HO sili r.r - X ) cos (y y ) 1 10 - 1 04t1) hZ

y i m H.x~, 1)) = 2 (4 H , sin (y .y) cos (s y ) (10-IOJC)

. , 7 nn

""9 Y ) = (;;:I H~ cos (F .) sin (y .), where y has the same expression as that given in Eq. (10-95) for TM modes.

Equation (10-96a) for cutoff frequency also applies here. For TE modes, either rrz or r 1 (but not both) can be zero. If a > h, the cutoff frequency is the l o w c ~ whcn nt = 1 and 11 = 0:

The corresponding cutoff wavelength is

Hence the TE,, mode is the dorninunt mode oj u rectangular wuueguide with a > b. Because the TE,, mode has the lowest attenuation of all modes in a rectangular

. waveguide and its electric field is definitely polarized in one direction everywhere, it is of particular practical importance.

10-4 RECTANGULAR WAVEGUIDES 473 ' I ' . ,

Example 10-7 (h) write the instantaneous field expressions for the TE,, mode in a rectangular waveguide having sides u and b. (b) Sketch the electric and magnetic field lines in typical;*-y. y-r, and x-r planeii(c1 Sketch the surface currents on the

- guide walls. . f

Solution l

a) The instantaneous field expressions for the dominant TE,, mode are obta~ned by multiplying the haso or expressions in Eqs. (10-103) and (10-104a) through (10-104d) with e"".'"' and then taking the real pan of the product. We have, for m = 1 and 11 = 0,

EJx, .,:, z ; t) = 0 (10-107a)

h) We scc from k ; r . 1 l ; ) IO72 j i l i r o~~g l~ ( 10 - 1 O?f ) i11;it [hi. TE, ,, mu<ii. ha> uni? three nonzero fie!d -omponcnts-namely. E,, H,, and Hz. In a typicnl I-) plane, say, when sin (or - p:) = 1, both E, and H, vary as sin j q i ; i ) and arc independent of y, as jhown in Fig. 10-l?(a). . .

ln a typical y-z plane, for example, at r = u/Z or sin (n.x/a) = 1 and cos (nx/a) = 0, we o~ily have E, and H,, both of which vary sinusoidally with QI. A sketch of E, a n 3 il, ;I[ i = 0 is givcn in Fig. 10-1?(b).

The sketch in an x-; plane will show all three nonzero field components - Ey. FI,. and 1-1;. Thc i l ~ p c of h e I1 lines : ~ t i = O is govcrncd by t l ~ a kiiIo\iinp equation:

$ ("1 tan (" s) tan p:,

which can be ~iscd to dciw thc H lincs in Fig. 10-i7(c). Thcc l ina ;lit ini1i.pr.n- dcllL 0I,jn.

e) The surface curreni ~ h s i t y an guide ivalls, J,, is related to the magnetic field intensity by Eq. (7-5Ob):

------ Magnetic field lines

Fig. 10-12 Field lines for TE,, mode in rectangular waveguide.

10-4. Wave. -

Fig. 10-13 Surface currents on guide walls for TE:, mode in rectangular waveguide.

. . .:.P Q .I \: . 1 0 4 RECTANGULAR WAVEGUIDES 475 -. . - 1 , . 4 - .. I . , . . . . .-

$ , , ' ? " . . : b '

. I * . . where a, is the outward normal to the will surface and H is the magnetic field . .

intensity at the &all. We have, at t = O7 * ,,'" . ' .. . ,' I

JS(x * 0) = - 4 N , ( O , y, z ; 0) = -a;&@ cos pr (10-Ilia)

J ~ ( x = a) = a,,h,(u, y, z ; 0) = J,(x = 0) (10-lllbj

J,(y -0) = a.H,(x,O,r;O) - n,H,(x, 0,z; 0)

(10-111c)

J,(Y = b) = -J,(y = 0). (10-1 1 Id) The surface currents on the inside walls at x = 0 and at y = h are sketched in Fig. 10-13.

10-4.3 Attenuation in ~ e c t a n ~ u l a r Waveguides

Allcnuntion for propagating modts rodits whm there are losbes in the dielec~ric and in the imperfectly conducting guide walls. I3ecause these losses are uscilly very snull, we will assume, a? In the casc of parallel-plate waveguides, that thc trnnsversc field patterns are not appreciably affected by the losses. The attenuation constant

,Y - due to losses in the dielxtric s:io be obtained by sobstitutin, - E + ( ~ , ' J ( : J J for c in Eq. (10-95). The result is exactly the same as that given in Eq. (10-751, which is rcpeated below: ,

where o and i i are the conductivity and intrinsic impedance of the dielectric m L' d ' lum respectively, and J', is given by Eq. (!0-96a1.

- i o determine the attenuation constant due to wall losses, we make use of Eq.

(10-77). The derivation:; of r, for the general TM.,,. and TE,.,, modes tend to be tedious. Below we obtain the formula for the dominant TE,, mode, which is the most important of all propagaiing modes in a rectangular waveguide.

For the TE,, mode the only nonzero field components are E,, H,, and Hz. Letting m = 1, n = 0. and h = (lr/uj in Eqs. (10-1046) and (10-104c), we calculate the time-average power i-lowing through a cross section of the waveguide:

In order to calculate the time-average power lost in the conducting walls per , . c ' . L

k , I

. . unit length. we must consider all four' waIls. From Eqs. (10-110), (10-1031, and . .- b

(10-104c) we see that q I

J:(.u = 0) = J:(x = a) = -a,HP(s = 0) = -a,Ho (10-114a) and

D

pa = ,, 0 s (2 .) - a. - H. sin (2 .). 7T

The total power loss is then double the sum of the losses in the walls at s = 0 and b

at y = 0. We have

and

Substitution of Eqs. (10-116a) and (10-116b) in Eq. (10-115) yields

The last expression is the result df recognizing that

Inserting Eqs. (10-1 13) and (10-1 17) in Eq. (10-77), we obtain

i14a)

114b)

j and

-115)

!16a)

r

1 l6b)

1 -1 17)

i-118)

.. -,. >~*-.*kp... &if A..'~+.,.'. '3.. .. . ** . - . -a- * 1 '.*--t*>- . . , 4 . R~CTANGUUR WAVEGUIDES 477

r. .* . . *'< . . "' P'"

"-'C ' : - - \ . I -

I .. : Equation (10-118) reveals a rather complicated dependence of (o,h, , on the +-,

ratio (M). It tends to infinity when f is ddse to the cutoff frequency, decrezses toward a m i n i m 6 as f increases, and increapes again steadily for further increases in f. v.: 7'

For a g i a n guide width u, the attenvatian decreases as b increases. However, increasing b also decreases the cutoff frequency of the next higher-order mode TE, (or TM,,), with the consequence that thc available bandwidth for the dominant TE,, mode (the range of frequencies over which TE,, is the only possible propagating mode) is reduced. The uiual compromise is td choose the ratio b/a in the neighborhood of +.

.I

Example 10-8 A TE,, wave at 10 (GHz) propagates in a brass - o, = 1.57 x 10' (S/m)-rectangular waveguide with inner dimensions a = 1.5 (cm) and b = . 0.6 (cm), which is filled with polyethylene-c, = 2.25, p, = I , loss tangent = 4 x lo-'. Determine (a) the phase constant, (b) the guide wavelength, (c) the phase velocity, (d) the wave impedance, (e) the attenuation constant due to loss in the dielectric, and (f) the attenuation constant due to loss in the guide walls.

Solution: At f' = 101° (Liz), the wavelength in unbounded polyethylene is

The cutoff frequency for the TE1, mode is, from Eq. (10-105),

U j ; = - - - 2 x l oH

- = 0.667 x 10" ( H z ) 2~ 2 x (1.5 x lo-')

a) The phase constant is, from Eq. (10-1 IS),

= 7 4 . 5 ~ = 234 (rad/m)

b) TIle guide wavelength is, from Eq. (10-34),

c ) The phase velocity is, irom Eq. (10-36),

. , - . t,.' ..' ,*,. 6 8 4 . . r t

"" -478 WAVEGUIDES AND CAVITY RESONATORS / 10

> < , . 8 , ., . a

d) The wave impedance is, from Eq. (10-47), .. . , . , , 2 - _ ' I .

t L -, 8 (~n),, = " JEi; 3 7 7 7 ~ . J = 0.745

= 337.4 (R).

e) The attenuation constant due to loss in dielectric is obtained from Eq. (10-112). The effective conductivity for polyethylene at 10 (GHz) can be determined from the given loss tangent by using Eq. (7-93):

= 5 x (Slm). Thus.

d C( - - Z T E =

5 1 0 - ~ d - 2 2

x 337.4 = 0.084 (Nplm)

f ) Theattenuation constant due to loss in Lhc guidc wdls is Ibittid Srvm Eq. ( (0-1 I ( ) ) . . - We have, from Eq. (9-26b). . --.

10-5 DIELECTRIC WAVEGUlDES

In previous sections we discussed the behavior of electromagnetic waves propagating , - along waveguides with conducting wails. We now show that dielectric slabs and rods

without conducting walls can also support guided-wave modes that are confined essentially within the dielectric medium.

Figure 10-14 shows a lon_gitudinai cross section of a dielectric-slab waveguide of thickness d. For simplicity we consider this a problem with no dependence on

€0, PO dielectric-slab wavepuide.

0-112). L'LI from

'.

i 0-1 19).

,--

';p In)

-B!m).

Fagating and rods confined

avcguidc ,&nce on

r-

.n G'

1 ! - i . . - % . ( 4 . . - - . 'L'.,,,. I;

the x coordinate. Let ed and9,ud be, respectively, the permittivity and permeability of the dielectric slab,.which is situated in free space (a,, p,). We assume that the dielectric is lossless and that waves propagate in the +z direction. The behavior of TM and ,

TE modes will now be halyzed separately. , ,

10-5.1 TM Waves alohg a Dlelectric Slab

For transvcrsc tnagnctic waves, M, = 0. Sincc thcrc is no x-dcpendcncc, Eq. (10 - 5 3 ) applies. We have

where 7 7

11- = y- + w2/J€.

Solutions of Eq. (10-120) must be considered in both the slab and the free-space regions, and they must b? matched at the boundaries.

In the slab region we Jssume that the wavcs propagate in thc +:direction without attenuation (lossless dielectric); that is, we assume

The solution of Eq. (10-120) in the dielestri~ slab may connin both a sine term and a cosine term. which are respectively an odd and an even ful~ction of y:

EP( y ) = E, sin k,y + E, cos k,y, d

lyj I - > 2. (10-123)

where

A: = &pdcd - lj2 = h J . (10-124)

In the free-space regions I y > d/2. and y < - d/2), the waves must decay exponentially so that they are guided along the slab and do not radiate away from it. We have

10.' - dl 2) d Y 2 - i (10-12ja) Lq( y) =

c , ~ z ( P + ~ / ~ ) re- ' ~ 1 5 - - ri 110-12jb) -- . 2' where

Equations (10-124) and (10-126) are called dispersion relations because they show the nonlinear dependence of the phase constant /3 on w.

At this stage we have not yet determined the values of k, and a; nor have we found the relationships among the amplitudes E,, E,, C,. and C,. In the following, we will consider the odd and even TM modes separately.

a ) Odd TM Modes. For odd TM modes, E:(!) is described by a sine function that is antisymmetric with respect to the J = 0 plane. The only other field components, E;(y) and I I : ( ~ ) , are obtained from Eqs. (10-23d) and (10-23a) respectively.

i) In the dielectric region, 1j.I I dl?: 41.

EP(y) = E , sin kyy

jB E:(y) = -- li, E, cos k,y (10-127b)

j*ed H;(y) = - E, cos k,y. (10-137~) - k?. - _ -. --. ii) In the upper free-space region, y >_ d/2:

where C, in Eq. (10-125a) has been set to cqual E , sin (k,d/2), which is the i d u e of Ez(y) in Eq. (10-127a) at the upper interface. y = d/2.

iii) In the lower free-space region. y I -d/2:

( k ; d ) e z O + d i I i E P ( ~ ) = - E, sin - (10-129a)

where C1 in Eq. (10-125b) has becn set to equal - E,, sin (kyd/2), which is the value of E:(y) in Eq. (10-127a) at the lower interface y = -42 . Now we must determine k , and T X for a given angular frequency of excitation

w. The continuity of H , at thi dielectric surface requires that Hl(d/2) computed

. . I I .

, . , ' . . .:. ' . ..q ?,. ., "." p... ' 1

2 .

avu we , . I . -

, , ' . '''i.~ , A - (Odd TM modes). (10-130) I .. /

- . ,

I By adding dispersion relations Eqs. (10-124) and (10- 126), we find

\

Equations (10-130) m d (10-131b) can be combined to glve an expression in '

which k, is thc only cnhno\vn:

Unfortunately 111.: tsanscendentri! quation. Eq. (10-132). caniiot be sol'\-ed analytically. But for ;L given o and given values of E, , p,, and ti of the dielectric slab, both the left and the right sidcs of Eq. (10-132) can be plotted versus k,.. The intersections of ..he two curves give the values of k , for odd Th4 modes. of which there are o h y 2 finite number, indicating that there are only a finite number of possible modes. This is in contrast with the infinite number of modes possible ir. waveguides with exlosed conducting walls.

We note from Eq. ( 1 0- 1 X'a) that EZ = 0 for !. = 0. Hence, a perfecdy son- (111cli1ip ~ I ; I I ~c m y ,>c ~IIII.O(III~:(YI to co i~ i~ , i [ l t : w i ~ l ~ ~ l i c 1, . ii pl.111~: ~ ~ ~ l / i t ) ( ~ t

I I : I I I I I I l o v I I I i i ; i 01' odii 1',U w a v o propagating along a dielectric-slab waveguide of thickness d are the same 2s those of the cor re~pocdin~ Th4 modes supported by a dielectric slab of a thickness dl2 that is backed by a perfectly conducting plane.

The surjuce impedance looking down from above on the surface or ciclectric slab is

El ;:,, = - -- - Y - j - (TM modes),

N,O we, (10-133)

482 '."WAVEGUIDES AND CAVITY RESONATORS / 10 ,. , ,

-

The other nonzero field components, E; and H:, both inside and outside the i . , dielectric slab can be obtained in exactiy the same manner as in the case of odd

I TM modes (see Problem P.lO-25). Instead of Eq. (10-130). the characteristic relation between and a now becomes

a -= -- k,d E0 cot -

7 (Even TM modes), (10-135)

1; ,. Ed - I I

which can be used in conjunction with Eq. (10-131b) to determine the transverse wavenumber k, and the transverse attenuation constant a. The several solutions

a \ correspond to the several even Thf modes that can exist in the dielectric slab waveguide of thickness d. Of course. in this case a conducting plane cunnot be placed at !: = 0 without disturbiw the whole field structure.

From Eqs. (10-124) and (10-126), it is easy to see that the phase constant, P, of propagating TM waves lies between the intrinsic phase constant of the free space, - -- k, = o J,LL,,E~. and that of the dielectric, k , = Q Jpdcd; that is.

As approaches the value of w&, Eq. (10-126) indicates that a approaches zero. An absence of attenuation means that tlic waves are no longer bound to the siab. The limiting frequencies under this condition are called the cutof frequencies of the dielectric waveguide. From Eq. (10-124) we have k,. = o, , /L~ ,c , - .poco at cutoff. Substitution into Eqs. (10-132) and (10-135) with cc set to zero yields the following relations for TM modes. At cut-off:

Odd TM Modes

tan r+ = 0

I I Even TM Modes I I o d I c o t ( + , / 3 ) = 0 I

It is seen that j,, = 0 for n = 1. This means that the lowest-order odd TM mode can propagate along a die!ectric-slab waveguide regardless of the thickness of the siab. As the frequency of a given T X I wave increases beyond the corresponding cutoff frequency, a increases and the wave clings more tightly to the slab.

. . , " . . , , . .-.. ,,.,.., " . .). . , . . . : : . , , . ;,.: i.!;,:, ; .:h:;

" . . : . :. - . ,. ,. ie the ' : : a .

, i

,f odd , . . . .: / :.

svcrse utions c slab lorbe .I

:. p, of space,

/?

s zero. 2 s. of the cutoff. iowing

Ad 7- ness of

.on2

. . 30-5.2 TE Waves along a Dielectric Slab

For transverse electric waves, Ez = 0, and Eq. (10-67) applies

wherc k, has been defined in Eq. (10-121). The solution for H:(y) may also contain both a sine term and a cosine term:

(1 H,;(Y) == H. Sill /i;J'-t- H e COS k,$, b.1 7. (10-138) &

In the free-space regions (,v > dl2 and :r < -d/2), the waves must decay exponentially We write

where u is defined in Eq. (10-126). Following the same procedure dr used for TLI waves, we consider the odd and even TE modes separately. Besides H;(J) . the only other field components are H:iy) and E:(),). which can be obtained from Eqi. I 10- 42b) and f 10-42c).

ii) In the upper free-<pace region, J- 2 (112:

..&..' " . . . *.A '.. . I . ,

"484 WAVEGUIDES AND CAVITY RESONATORS 1 10 *

iii) In the lower free-space region, y I - 4 2 :

- , ~ : ( y ) = - (H. sin y) 8 ( Y t d 1 2 ) ' (10-141a)

A relation between k, and a can be obtained by equating EP(y). given in '., Eqs. (10-139d and (10-140c), at J. = d / 2 . Thus,

' l o tan - (Odd TE modes),

which is seen to be closely analogous to the characteristic equation, Eq. (10- 1301, for odd TM modes. Equations (10-131b) and (10-147) can- bcqmbined in the manner of Eq. (10-132) to find k,. graphically. Froin li,.. x can bc found from Eq. (10-131b).

From a position of looking down from above. the surface impedance of the dielectric slab is

EO - -IT- - -.(J'Po H: IT (TE modes). (10-143)

which is a c:~p;~citivc rc:lct;~ncc. Tlcnct:. ir TE .srir./irc.r. w t r r v cetrrl lw . s ~ r p p ~ r . / , ~ r l /)I* rr

c . u p ~ c . i / I ' r . ~ , S I I I ~ \ I I ~ ~ P .

b) Even TE Modes. For even TE modes, is described by a cosine function that is symmetric with respect to the y = 0 plane.

H;(.V) = H,, cos k,.!., 1j.l 5 dl?. (10-144)

The other nonzero field components, H: and E,O, both inside and outside the dielectric slab can be obtained in the same manner as for odd TE modes (see Problem P.lO-27). The characteristic relation between k,. and a is ciosely analogous to that for even TM modes as given in Eq. (10-135):

1t is easy to see that the expressions for the cutolf frequencies given in Eqs. (10-136a, b) apply also to TE modes. The characteristic relations for all the propagating modes along a dielectric-slab waveguide of a thickness d are listed in Table 10 -2.

141 b)

-141~)

ven in

'. 1-142)

-130). n the lrom /-

of the

3- 143)

~d by u

.nctlon

(J- 144)

ide the x s (see , tln~11-

0 145)

f--

.n Efl.; ,111

:e listed

Example 10-9 A dielectric-slab waveguide with constitutive parameters p, = p,, and ed = 3 . 5 0 ~ ~ is situated in free space. Determ~ne the minimum thickness of the slab so that a TIM br TE wave of the even lype at a frequency 20 GHz may propagate along the guide.

Solution: The lowest T!/! and TE waws of the even type have the same cutoff frequency along a dielectric-slab waveguide:

Therefore, C

d",," = -- 7-

Example 10-10 (a) Obtain an :~pproximatc espression Sor the decaying r;ltc: of the ~ Io~n i~ i :~n l TM SLII~:ICC w.ivc oi~is i~lc 0 U il very ~ h i n ~liclcctric-s1:1\> \v:~vquidc. ( b j

Find the time-a;.ciage pow: per unit width transmitted in the transverse direction along the guide.

Solution

a) Tile dominant TM wave is the odd mode having a zero cutoff frequency- f,, = 0 for n = 1, independent of the slab thickness (see Table 10-2). With a slab

. . < , . .-. , 'dk".486 ' I 3)

. . ' . , WAVEGUIDES AND CAVITY RESONATORS 1 10

I , .... - . . i - that is very thin compared to the operating wavelength; k,,d/2 << 1, tan (kyd/2)

z k#2, and Eq. (10-130) bepmes .

Z a s - k , d . €0 2 (10-146) 2%

Using Eq. (10-131a), Eq. (10-146) can be written approximately as

In Eq. (10-147), it h i s been assumed that rd/2 << ed/eO

?. b) The time-average Poynting vector in the +: direction in the dielectric slab is

Pa, = $h(- aYEy x a,H,).

Using Eqs. (10-127b) and (10-117~). we have P,,,. = n,P,,,:and

(1; 2 Pa. = 2 J' iP,, d!. = ,

0 cos' ( k ~ ) dy

k,

where

and

In this section we have studied the characteristics of TM and TE waves guided by dielectric slabs. The same principles govern the transmission of light waves along round quartz fibers that form optical waveguides. Optical waveguides are of great importance as transmission media for communication systems because of their low- loss and large-bandwidth properties. Their analysis requires the knowledge of Bessel functions which we do not assume in this book.

10-6 CAVITY RESONATORS

We have previously pointed out that at UHF (300 MHz to 3 GHz) and higher frequencies, ordinary lumped-circuit elements such as R. L, and C are difficult to make. and stray fields become important. Circuits with dimensions comparable to the opcr:lt ing w:lvclcnplh hccomc cllicicnr r:~tli:l tors and will interfere with othcr circ~~ifs and systems. Furthermore, corivcritiolial wire circuits tend to have a high elYcctive resistance both because of energy loss through radiation and as a result of skin effect. To provide a resonant circuit at UHF and higher frequencies, we look to an enclosure (a ca\,ity) completely surrounded by conducting walls. Such a shielded enclosure confines electromagnetic lields inside and furnishes large areas for current flow. thus eliminating radiation and high-resistance effects. These enclosures have natural

Long great

:r low- Basel

.: :r f:e- nuhe. io the . . ":cJlts kc:{- . ciiect. 2:os - i0:;UrZ : How, xtural

(a) Probe c.sci:at~on. (b) Loop excitation.

Fig. 10-13 Excitation of cavity modcs by a coaxial line.

rcsonant frequencies anc: it vcry high Q tquitlity factor), i ~nd itre callcd ~ . i l c i r j ~ r C . w

nutors. In this section WL will.study the properties of rectangular cavity resonators. Consider a rec tanghr waveguide with both ends c!osed by a conductin. wiill.

Thc interior dimensions 01' tllc ci~vity ;iru n, b, ;~nd (1. its shown i n frig I0 - 15. Sinm both TM and TE modei can exist in a rectangular guide, we expect Tkl and TE modes in a rectdn:u~ar resonator too. However, the designation of TM and TE modes in a resonator is .iot m i p r because we are free to choose i or y or r as the "direction of propap~ti~m": that is, tllere is no unique "longitudinal direction." For example, a TE mode with respect to the : axis could be a TM mode with rzspect t o the y axis.

For our purposes. we cbnorr the : uxis us rbr rejetwice "direction of propaq3tion." In actuality, the existent: of conuucting end walls at .- = 0 and r = d gives rise to multiple reflections and bets up standing waves; no wave propagates in an enclosed cavity. A three-symbol (~nnp) subscript is needed to designate a TIM or TE standing-wave pattern in a cavity resonator.

10-6.1 TM,,,, Modes

The expressions for the rransverse variations of the field components for TM,,, modes in ii iv:ivspide l s~vc hccn givcn in E q s ( 10--92) ilntl ( 1 ii %I:I, b. c. ~ 1 ) . h ~ t c that the longiti1dik11 v:inilios lor 2 waic traveling in the +: direction is dcscribeil 1.. ",, -1 facLOr e-:'= Or r-;,:': . as indichted in Eq. (10-54). This wave will be reflected by the end wall at .- = d ; and ihe reflected wave, going in the -: direction. is described by a factor rjD'. he si~perposition of a term with r - I#z and another of the same amplitudet with r"' resu!is in a standin: wave of the sin pi or cos P; type. Which should it be'? The answer io this question depends on the particular field component.

' The reflection coefficient at a lrrfcct cocductor is - I .

3 .'.-,. . I ., , ' . . 488 WAVEGUIDES AND C A V I N RESONATORS 1 10

Consider the transverse component E,(x, y, z). Boundary conditions at the . conducting surfaces require that it be zero at z = 0 and z = d. This means that (1)

its ,--dependence be of the sin Br type and that (2) P = pnld. The same argument applies to the other transverse electric field component E,(x, y; z) .

Recalling that the appearance of the factor (- y) in Eqs. (10-94a) and (10-94b) is the result of a differentiation with respect to z, we conclude that the other components E,(x, y, z), H,(x, y, z), and H,(x, y, z), which do not contain the factor (-;)), dust vary according to cos /3z. We have then. from Eqs. (10-92) and (10-94a, b. c, d). the following p11asor.s of thc ficld components for TM,ll,lp modcs ill a rcct;~npl.~r cavity resonator.

~ ( x . y, 2) = E~ sin r: x) sin (y y) cos (7 z)

H,(x, y, 2) = -- "'(7) 12 - Eo cos > z ) ) - x sin - v cos ("$;), (10-149e)

where

From Eq. (10-95), we obtain the following expression for the resonant frequency for TM,,, modes:

f m n p 2 t

I

10-5.2 TE,,, Modes

For TE,,, modes ( E , = 0). the phasor expressions for the standing-wave field com- .. . ponents can be written from Eqs. (10-103) and (10-104a, b, c, d). We follow the same

rules as those we used for TM,,, modes; namely, (1) the transverse (tangential) electric field components must vanish at z = 0 and z = d, and (2) the factory indicates

m :

st the

:r com- ~r ('v), b. c, d), m&u

1-149a) i;.

J-149b)

I-149~)

n

J-1-, -)

O-l4e)

0-149f)

quency

,lo-i50)

/--.

. d J iI-

:nc same qent la l ) indicates

.. : . . . : .' . ,._ . .. , ,!., ,. , ,... :. ,!-, . .:;!. . (.' . . . ' ,( ": ' '.,'," ' -7 . . !; ? f . ... . . .. : . . ',

. . . . . . . , , , , ; , .. '. ' ...; I ! . . . . . . ' . ,: r . .. . ,

\ L. -. . f

+ 1 . . ' ' 10-6 [ CAVITY RESONATORS 489 .# ,: % . <

i I

a ncgative partid .difkrcntialion with rcspcct to z. Thc first rulc rcquircs ;l sin (p.rr:/tl) factor in E,(x; y; z).knd Ey(x, y, z), as well in H,(x, y,z); and the second rule

. ' indicates a cos (&Id) factor in H,(x, y,:,'z) and Hy(x , y, z), and the replacement of y

a by -(psl /d) . ~ h u s ,

jup (y) Ho cos (y x) sin (y y) sin (7 :) EJx, Y , 4 = - h2

Ey(x, y, Z) = ' ) H sin :) 0 s ( ) sin ) (1 0- 1 5 i c) h2 \ a

where I ? h a s been given in Eq. (10-143f). The expression for resonant frequency. f,,,, remains the s ahe ;IS that obtained for TM,,, modes in Eq. (10-150). Different modes having the same resonant frequency are called dcgenrrate rnorles. The mode with the lowest resonant frequency for a given cavity size is referred to as the riorni~lanr mode.

A particular mode n a cavity resonaror.(or a waveguide) may be txcited from a coaxial line by means of a small probz or loop antenna. In Fig. 10-13a) a probe is shown that is the tip of the inner conductor of a coaxial cable and protrudes into a cavity at a location whcrc thc clectric ficld is a maximum for thc dcsircd modc. Tiic probe is, in fact, an anttnna that coupies electromagnetic energy into the resou~tar. Alternatively, a cavity r:sonator may be excited through thz introduction of a small loop at a piace where the magnetic flux of the desired mode linking the lcop is a maximum. Figure 10-15(b) illustrates such an arrangement. Of course, the source frequency from the coa ikll line must be the same as the rcsonant frequency of the desired mode in the cavlty.

AS an example, for the TE,,, mode in an a x b x d rectangular cavity, there are only three nonzero held componznts :

. j ( ! ) l l ~ E,. = -4- I!,, sin

11-'1

.. --. 7i -

Fix = -- H,, sin l r a d

H= = H, cos !: s) sin /: :) ,

This mode may be exclted by a probe inserted in the center region of the top or bottom face where E,, is maximum, as shown in Fig. 10-15a, or by a loop to couple

_ . , , a maximum H, placed inside the front or back face, as shown in Fig. 10-15b. The 4 , . , best location of a probe or a loop is affected by the impedance-matching requirements

of the microwave circuit of which the resonator is a part. 'J ci

A commonly used method for coupling energy from a waveguide to a cavity resonator is the introduction of a hole or iris at an appropriate location in the cavity wall. The field in the waveguide at the hole must have a component that is favorable in exciting the desired mode in the resonator.

. . Example 10-11 Determine the dominant modes and their frequencies in an air- filled rectangular cavity resonator for (a) a > b > d. (b) n > d > b, and (c) a = b = d.

a:. where a. b, and d are the dimensions in the .u, y, and z directions respectively.

Solutiou: With the z axis chosen as .the reference "direction of propagation": First, for TM,,,,, modes. Eqs. (10-14% b, c, d, c) show that ncirhcr 111 nor r r can bl; zero, but that p can be zero; second, for TE,,,,,, modes, Eqs. (10-151~1, b, c, d. e) show that either nr or 11 (1x11 not both rn xnd 11) can hc zcro. hut that p cinnot be zcro. Thus. the modes of the lowest orders arc

I--..

TM,,,, TEoll. and TElo, .

The resonant frequency for both TM and?^ modes is given by Eq. (10-150).

a) For a > h > d: Thc lowest resonant Srcqucncy is

where c is the velocity of light in free space. Therefore TM, ,, is the dominant modc.

1

b) For u > ti > b : The iowcst rcsonant frcqucncy is

i

and TE,,, is the dominant mode. i

c) For a = b = d, all three of the lowest-order modes (namely, TM,,,, TE,,,, and i

TE, ,,,) have the same ficld patterns. The resonant frequency of these degenerate i 1

modcs is !

I 10-6.3 Quality Factor of Cavity Resonator

A cavity resonator stores energy ir? the electric and magnetic fields for any particular mode pattern. In any practical cavity the walls have a finite conductivity; that is, a nonzero surface resistance. and the resulting powcr loss causes a decay of the stored i

. The Gents A

; <;. . :: a+-ytt,- -< -.. avity i-able

-I- air- ' = d ;

'.\ . . First, zzro, that

Y, the

,-

50).

inant

. and ::rate

/--

:cular i IS,

.tored

i

energy. The quality fdcror, or Q, of a resonator, like that of any resonant circuit. is a measure of the bandwidth of the resonator and is defined as

1 lme-average energy stored at a resonant frequency Q = 2n . (10-153) Energy dissipated in one period of this freauencv

(Dimensionless) . +

Let W be the total time-average energy in a cavity resonator. We write

where We and Wn, dencte the energes stored in the electric and magnetic field, respertively. If P, is the .ime-average power dissipated in the cav~ty, then the energy dissipated in one period li P, diridsd by frequency, and Eq. (10-1 53) can be wr~rten i s

In determining the Q of .I cavity at a resonant frequency, i r is cus:omary to nssume that the loss is smail c n o q h to allow the use of the field patterns without loss.

We will now find thi Q of an n x b x d cavity for the TE,,, mode that has three nonzero field compoaenrs $ivm in Eqs. (10-1524 b. and c). The time-average stored electric energy is

where we have used / I - ' := ja,n)' from Eq. (10-1491). The total time-a~ernge stored magnetic energy is

- - - Po ff; ' 7 ' ) - b I:) - + (!: 7 , h (?$I - =- b + I . (10-l56b) 4 d - ( 2 ,

From Eq. (10-150), the resonant frequency for the TE,,, mode is I 5 ? .

f l o l = - (10-157) \;& 4.

Substitution of fl0, from Eq. (10-157) in Eq. (10-156a) proves that at the resonant frequency We = W,. Thus,

w=2w,=2wm=- 8

(10-158)

To find P,, we note that the power loss unit area is 'i

.you = f (J, /~R, = ) IH,~~R, , (10-159)

where ~ H , I denotes the magnitude of the tangential component of the magnetic field at the cavity w;~lls. The pnwcr l o t i l l the : -- (1 (h:~chl a ; ~ l l is tlic .;:line ; I S t11:il .ill 1I1c 1 = 0 (front) wall. Similarly, the power loss i l l ille r = (Icfl) wall is tlm s;1111c ;IS

t h i in ihc Y = 0 (riphll ~v:~ll: i ~ n d ~ l ic lxn\cl- Inst ill tllc 1. - 1) ( t ~ ~ ~ , ~ c r ) ill is I I I C >:\tl~c as that in the y = 0 (lowcr) wall. LVc ha\e

1. 1.

PL = $ ~ a . ds = R,{J: J; IH.(Z = o)12 b d l + J'Sb JH_(X = o)l2 dy dz 0 0

-+ Sod S; iH,12 dx d~ i kd J; I H = J ~ d~ d~

- --

Using Eqs. (10-158) and (10- 160) In Eq. (10-155). we obtain T -. ,- ' . .'

~ f l o l ~ o a b d ( a ~ + d2) r -

Q l o l = (For TE,,, mode), (10-161) A I . -a * I - - - ~ , [ 2 b ( a ~ + d3) + nd(a2 + d2)] 1,- -. . - i e

where Jlo, 1x1s been given in Eq. (10-157).

Example 10-12 (a) What should be the size of a hollow cubic cavity made of copper s 'I

in order for it to have a dominant resonant frequency of 10 (GHz)? (b) Find the Q at that frcqucncy.

'Solution ,

a) For a cubic cavity, a = b = d: From Example 10-11, we know that TM,,,, 9

TEol 1, and TElol are degenerate dominant modes having the same field pat- I

terns. and that 3 x lo8

f i o l = - - - 101° (HZ). 42'2

1:rt in the s ; m e as :he same

of copper -1nd the Q

b) T h e expression of Q in Eq. (10-161) for a cubic cavity reduces t o

xi1 0 I P d ~ = f J- 0101 =

3R, ' i i i a l ~ o o . (10-162)

For copper, o = 5.80 x lo7 (S/m), we have

The Q ola cavity resonator is. thus. cstremcly high comparcd with that obt;linable liom lumped L-C resonant circuits. In practice, t h e preceding value is sornen hat lower d u e to losse: through feed connections a n d surface irregularities.

REVIEW QUESTIONS

R.10-I Why ar? the conmon typcs of ir.~nsmisiion lines not useful ior the long-distance sional trrm.mission of n~icrowiiv,: frequencies in the TEbI mode'! .

- R.10-2 What is memt by a a ~ r o f l ' n / l . q u n i c ~ of awaveguide?

RJO-3 Why 2re lumped-panmeter elements connected by wires not useiul as resonant c~rcuits at microwave frequencies?

R.10-4 What is the govrrning equation for slectric m d magnetic field intensity phasors in the dielectric region of a sttaight waveguide with a uniform cross section'?

R-10-5 What -re the rhrze basic types of propagating waves in a uniform waveguide?

R.10-6 Define wuce inlpt dance.

R.10-7 Explain why single-conductor hollow or dielectric-filled waveguides cannot support TEM waves.

R.10-8 Discuss the analytical procedure for studying the chamcteristics of Tb1 waves in a waveguide.

R.10-9 Discuss the rna1ytic:ll procedure ior studying the chamiteniiiis "f TE u;liei in i w;lvcgllltlc. ..

R.10-I0 What are e i q o i c ~ l ! ~ . ~ of a boundair-raise problem?

8.10-11 Can a waveguide have more than one cutoff frequency? On what factors does the cutoff frequency of a waveguiae depend.

' R.10-I3 Is the guide wavelength of a propagating wave in a waveguide longer or shorter than the wavelength in the corresponding unbounded dielectric medium?

R.lO-14 In what way does the wave impedance in a waveguide depend on frequency:

a) For a propagating TEM wave? b) For a propagating T M wave? c) For a propagating TE wave'?

R.lO-15 What is the significance of a purely reactive wave impedance?

R.10-16 Can one tell from an w-p diagram whether a certain propagating mode in a waveguide is dispersive? Explain.

R.10-17 Explain how one determines the phase velocity and the group velocity of a propagating mode from ~ t s w-/j' diagram.

R.10-18 What is meant by an eigenmode?

R.10-19 On what factors does the cutoff frequency of a parallel-plate waveguide depend?

R.lO-20 What is meant by the dominant mode of a waveguide? What is the dominant mode of :I parallci-plate wnvcguide? -I-.

R.10-21 Can a TM or TE wave with a wavelength 3 (cm) propagate in a parallel-plate waveguide whose plate separation is 1 (cm)? 2 (cm)? Explain.

R.10-22 Compare the cutol-l' frequencies of TM,, TM,, TM,, (m > n), and TE, modes in a parallel-plate waveguide.

R.lO-23 Does the attenuation constant due to dielectric losses increase or decrease with frequency.for TM and TE modes in a parallel-plate waveguide'?

R.10-24 Discuss the essential differecces in the frequency behavior of the attenuation caused by finite plate conductivity in a parallel-plate waveguide for TEM. TM, and TE modes.

R.10-25 State thc boundary conditions to bc satislicd by E, for TM wavcs in n rectangular waveguide.

R.lO-26 Which TM mode has the lowest cutoff frequency of all the TM modes in a rectangular waveguide?

K.10-27 State the boundafi conditions to bc satislicd by H, for TE waves in a rectangular waveguide.

R.lO-23 Which mode is the dominant mcde in a rectangular waveguide if (a) o > h, (b) n < h. and (c) a = h?

R.10-29 What is thc cu tor wavelength of the TE, , mode in a rectangular wavcguiuc?

R.10-30 Which are the nonzero ficld componcnts for the TE,, mode in n rectangular waveguide? - R.10-31 Discuss the general attenuation bchavior caused by wall losses as a function of frequency for the TE,, mode in a rectangular waveguide.

mode

r- I"UIC

R.lO-32 Discuss the factors that affect the choice of the linear dimensions n and b for the cross section of a rectangular paveguide. . . . . R.10-33 Why is it necessary that the permittivity d t h e dielectric slab in a dielectric waveguide

. . be larger than that Of the surrounding medium?

R.10-34 What are dispersion relations?

R.10-35 Can a dielectric-slab waveguide support an infinite number of discrete TM and TE modes? Explain.

R.lO-36 What kind of shrfxe can support a TM surfnce wave? A TE surhce wave?

R.10-37 What is the dominant, mode in a dielectric-slab waveguide? What is its cutodfrequency '?

R.10-38 Does the attenuation of the waves outside a dielectric slab waveguide increase or . decrease with slab thlckhess?

R.10-39 What are cavity resonators'? What are their most desirable properties'!

R.10-40 Are the field patterns in a cavity resonator traveling waves or standing waves'? How do they differ from those in .I waveguide'!

<

R.10- 11 In terms of field pmerns what does the TM, ,, mode signify? The TE, 2, mode?

R.10-42 What is the e%preision f a the resonmt frequency of TM,,,,,, modes in n rectanjulx cavity resonator of dimensicns a x h x d? Of TE,",,, modes'! .

R.10-43 Whar is meant by te~~clrel.c~te t~lodes? - k.10-4 What are the mod:s of the lowest orders in a rectangular cavity resonator?

R.10-4 Define the quality factor, Q, of a resonator.

R.10-46 Explain why the measured Q of a cavity resonator is lower ~ h a n rhe calculated vnius

PROBLEMS

P.10-1 Starting from the iwo time-harmonic Maxwell's curl equations in cylindrical zoordi- nates, Eqs. (7-8ja) and ( 7 4 5b), express the transverse field components E,, E,, H,, and H, in terms of the longitudinal corlponents E, and H,. What equations must E: and H, satlsfy'?

P.10-2 In studying the wale behavior in a straight waveguide having a uniform bur arbitrary cross section, it is txpedient :o find general formulas expressing the transverse field components in terr~is oftheir longitudinal components. We write

1.

* " , . \ i . # . . ,

where the subscript T denotes "transverse." Prove the followi& relatipns for time-harmonic excitation: -

1 a) Er = -p ( y V T E , - a , j q x VTHz) (10-163a)

where it2 is that given in Eq. (10-13).

P.lO-3 For rectangular waveguides. *'.

a) plot the universal circle diagrams relating tr, /~r and P / k versus],jj, b) plot the universal graphs of u h , . B/k, and ;.,,lib versus fjf,, C) find u,j11, u,/u, P / k , and i.,iiL ' ~ t 4 = 1.251,.

P.10-J Skctch thc c t ~ - - P tliugr;~m of 3 p:~rallcl-pl:~tc waveguide separated by 3 dielect~ic slab of thickness 6 and constitutive pa~unc te rs ( E . p ) for TbI ,. Th4,. and Tb13 modcs. Discuss

a) how b and the constitutive parameters affect the diagram. b) whether the same curves apply to TE modes.

- ----..

P.lO-5 Obtain the expressions for the surface charge density and the surface current density for TM, modes on tllc conducting piatcs o f ;I pa~~l ie i -ph tc w;~vcg~~iclc. Do the currcnts on the two plates flow in the same direction or in opposite directions?

P.lO-6 Obtain the expressions for the surface current density for TE, modes on the conducting plates of a parallel-plate waveguide. Do the currents on the two plates flow in the same direction or ii' opposite directions'?

P.lO-7 Sketch the electric and magnetjc field !ines for (a) the TM2 modc and (b) the TE, modc in a parallel-plate waveguide.

P.10-8 A waveguide is formed by two parallel copper sheets-0, = 5.80 x 10' (S/m)-separated by a 54cm) thick lossy Jiclectric - E , = 7.25, ii, = 1. a = 10- l o (S/nl). For :in operllting frequency of 10 (GHz), find /I, z,,, qC, up, u,, and i., for (a) thc TEM modc. (b) the T M , modc. and (c) the TM, mode.

P.10-9 Repeat problem P.10-8 for (a) the TE, mode and (b) the TE, mode.

P.lO-10 For a parallel-plate waveguide,

a) find he frequency (in tel'iils of the cutoff frequency 1;) at which the attenuation constant due to conductor losses for the TM, mode is a minimum,

b) obrzin thc ror~nulil for this mini~ri~~rn * ~ ~ t ~ : n u ; ~ l i o n constant. c) calculate this minimum 2, h r the T M , nwdc it' ~ l l c p:~r;dlcl plalcs arc niadc ol'coppcr

and spaced 5 (cm) apart in air.

P.lO-11 A parallel-plate waveguide made of two perfectly conduciing infinite planes spaced 3 (cm) apart in air operates at ;1 frequency I0 (GHz). Find the maximum time-average power that can bc propagltcd pcr unit width of the p i t i c without o voltagc hrcnkdown for

a) the TEM rnodc. !I) thc ' fM , nlotlc, c) tllc 'l.ll, I I I C ~ C .

ipsct..' power

PROBLEMS 497 ,

P.lO-12 Prove that the foflowing wavclcngth rclation holds for n uniform wavcguide: . ;- '% & ? I . I

where i., = guide wavelength, i = wavelength in unbounded dielectric medium, and i, = u/f, = cutoff wavelength.

P.10-13 For an (1 x b re~tas~gular waveguide operating at the T M , , mode.

a) derive the exprcssicns for the surfacc current dcnsitics on the conducting walls, 11) skctch illc surfacc c ~ ~ r r c n ~ s un Lhc walls at s ;= 13 and at ) = 6.

P.10-1.1 Calcuiare and list i 1 ascsndiny order the cutoff ticqucncics (in icmms ol' the cutolf frcqucncy of the dominant midz) of an u x b rcc~angular waveguidt: for tht Ijl!owlng modes: TE,,, T E l , 7 T E l , , T E , 2 . TE c: T M , TM12 . :mi TM,? [a) i f a - 3h. anti ~ h ) i T t ~ = h

1'.10-15 An air-lillsd ' r x b I ! , < .I < 36) rec~angular wavsguidz 1s to b\: consrrucred to opcmrc at 3 ICHz) irl the dornin;ill~ i:~oclc. :I/c clcsirc t l ~ opmrting Srcqtlcncy 10 hc ;it 1e:lat 20'1; h~ghcr Lhall l l ~ c culoif lrcqucncy ol't.ic c lu~nina~i~ nlodc .1nd :dso at lens^ 20:'" below ~112 cutotr rrquencq of the next higher-ordcr mod-.

a) Give a typicril dtsiy ; for the dimensions u and b. b) Calculate for your d-sign 1, I(,,, i.,, and .he ~c'.tve impedance at thi opcratlng frequency.

P.lO-16 Calculate and cornrxs the values of /, u,, u,, i,, :lnd ZT,,,, for .t 2.5 (cm) x 1.5 (cm) rcctangular wavcgu~tlc opcratng at 7.5 ICHz)

P) ~f the Lvavegu~de 1s lioilow. h) if thc waveguide is t~lled with .I d~electnc mcdiurn chnractcrizcd by E, = 2. p, = 1 2nd

a = 0.

P.10-17 An air-filled rcctangular wavcguide made oi copper and having transverse dimensions u = 7.20 (cm) and b = 3.40 (cm) operates at a frequency 3 (GHz) in the dominant mode. Find (a) fc, (b) i.,, (c) a,, and (d) the distance over which the field intensities of the propagating wave will be attenuated by 50%.

P.10-18 An avcrdgc ppowcr o, 1 ( kW) at I0 (GHz) is to be delivered to an antenna a: the TE,, mode by an air-filled rectangular copper waveguide l (m) long and having sides o = 1.35 (cm) and h = I .OO (cm). Finti

P.10-19 Find the maximum amount of 10-(GHz) hverage power that can be transmitted through an air-filled rectangular wavegl~de-u = 2.25 (cm), b = 1 .OO (cm)-at the TE,, node without .I

breakdown.

P.10-20 Determine the value,of (fl') a t which the attenuation constant due to conductor losses in an a x b rectangular waveguide for the TE;, mbdeis a minimum

". P.lO-21 Find the formula for the attenuation constant due to conductor losses in an a x b rectangular waveguide for the TM , , mode.

P.lO-22 Show that electromagnetic waves propagate along a dielectric waveguide with a velocity between that of plane-wave propagation in the dielectric medium and that in the medium outside.

P.10-23 Find the solutions of Eq. (10-132) for k, by plottmg qd versus k,.tI for d = 1 (cm) and E, = 3.25 if (a) f = 200 (MHz) and (b) f = 500 (MHz). Determine /3 and a for the lowest-order

i. odd TM modes at the two frequencies.

P.10-21 Repeat problem P.lO-23 using Eq. (10-135) for the lowest-order even TM modes.

P.lO-25 For an intinite dielectric-slab waveguide of thickness d situated in air, obtain the instantaneous expressions of all the nonzero field components for even TM modes in the slab, as well as in the upper and lower free-space regions.

P.lO-26 When the slab thickness of a dielectric-slab waveguide is very small in terms of the operating wavelength, the ficld intensities decay very slowly away from the-slab --_ surface. and the propagation constant is nearly equal to that of the surrounding medium.

a) Show that if k,d <c 1, the following relations hold approximately for the dominant TE rnodc:

11 2 k,,

I '

where k , = w,/pded and k , = o,,,%,. h) For a slab of thickncss 5 (mm) and dielectric constant 3. estimatc the distance from the

slab surface at which the ficld intensities have dccaycd to 30.8% of their valucs at ~ h c surface for an operating frequency of 300 (MHz).

P.lO-27 For an infinite dielectric-slab waveguide of thickness d situated in free space, obtain the instantaneous expressions of all the nonzero field components for even TE modes in the slab, as well as In the upper and lower free-space regions. Derive Eq. (10-145).

P.10-28 A waveguide consists of a ~ i inlin~tc diclcctr~c slab (ti. 11,) ol' tllickncss ti 111at is sitting on a perfect conductor. .

3) What are the propagating modes and what are their cutoff frequencies? b) Obtain the phasor expressions for the surface current and surface charge densities on

the conducting base for the propagating modes.

P.lO-29 Given an air-filled lossless rectangular cavity resonator with dimensions 8 Icm) x 6 (cm) x 5 (cm). find the first twelve lowest-order modes and their resonant frequencies.

P.lO-30 An air-filled rectangular cavity with brass walls-E". po, a = 1.57 x lo7 (S/m)--has the following dimensions: a = 4 (cm), b = 3 (cm). and ti = 5 (cm).

a) Determine the dominant mode and its resonant frequency for this cavity. b) Find the Q and the time-average stored electric and magnetic energies at the resonant

frequency, assuming H , to be 0.1 (Aim).

-- .... ... %;$ w. ,.: i; , ,_ . (_ i . :@@y$,.?!y$; ' ...F y% Y;:y+y; - . . ,,. ~7&z.w.::,$y'c~? ,; - : ::,,; , , <;.* '-, .. , , v .

..1. :. ... . . . . . . . ;,:. -:; ,&.',, .,::;,;;, . . . . . . . . . . . . ., ., .A,. ,':., .... . . . .+he :. :., . ; ". ' .' , .' . , . < - . >..?>.........+..I: . . . . . . . \., , i ,: . . . . . . . I . , . . . . . . f . ' '.

. . . . . .

. , , .. .. . . . ... +,, -: ;; ,,,,t" ,, .,, +. $ : ' ~ : ; : ~ " . . : ~ . & ~ ~ & , ~ , . ~ , . " . c ~ ~ ~ . ~ ~n?4~r~ l i - . ~ - ' ++ '4y . :+~

. . . . . . ,!". , . PROBLEMS '499 . ' , ' '

. . . . . ,. . ,

:ctor losses 1 ' 1

I (an) and )west-order ..

modes.

.Ice from the

.:;llucs at the

pacc, obtain :s in he slab,

the resonant

P.lO-31 If the rectangular cavity in Problem P.10-30 is filled with a lossless dielectric matenal having a dielectric constant 2.5, find

a) the resonant frequency of tht: dominant mode, b) the Q,

I c) the time-aver:\ge stored clcctric and magnetic energies at the rcsonanr frequency, assuming II,, lo bc O.1 (A/m). -..

P.lO-32 For an air-filled rectangular cavity resonator,

a) calculate its Q for the TE,,,, mode if its dimensions are u = d = I.Sh, I)) dcicrminc how much h \llould bc rncrcascd In ordcr to make Q 70Zhighcr.

P.10-33 Derive an expression for the Q of an air-filled a x b x d rectangular resohator for the TM , , , mode.

Fig. 10-16 A ring-shaped resonator with a narrow center part (Problem P.lO-34).

P.10-34 In some microwave applicsrions ring-ihaped cavity resonators with II vzry narrow center part are used. A c r x s section ofsuch a resonator is shown in Fig. 10-16. in which d is Lery small compared with the resonimt waveiengrh. Assuming that this resonator can be represented approximately by a purallsl cotnhination of the capacitance of the n:lrrow center parr and the inductance of the rest of :he structure, find

a ) thc ;~pproxim;:t : rcsonant I'rcqucncy. b) the appro xi mat^: resonant wavelength.

11-1 INTRODUCTION

In Cllaptcr 8 we studied the propag:ltion chnractcristics of plane electromitgnetic wavcs i i i sourcc-(roc ~ilcrlia witliout co~~sidcri~i$ ho\v tlic W:IVCS \VCI.C ~ C I I C ~ : I ( C ~ . OI' course. the waves must origina~c I'rotii sourccs, which in clcctrolna~~~ctic terms ;~ rc time-varying charges and currents. In order to radiate e l e ~ t r ~ m a ~ n e t i c e n e r g efficiently in prescribed directions, the charges and currents must be distributed in specific ways. Airtennns are structures designed for radiating electromagnetic energy effectively in a prescribed manncr. Without an eficient antcnna, clectrom~~gnctic energy would be localized, and wireless transmission of information over long distances would be impossible.

An antenna may be a single straight wire or a conducting loop excited by a voltase source, an aperture at the end of a waveguide, or a complex array of these properly arranged radiating elements. Reflectors and lenses may be used to accentuate certain radiation characteristics. Among radiation characteristics of importance are field pattern, directivity, impedance, and bandwidth. These parameters\,will be examined when particular antenna types are studied in this chapter.

To study electromagnetic radiation we must call upon our knowledge of Max- well's equations and relate electric' and magnetic fields to time-varying charge and current distributions. A primary difficulty of this task is that the charge and current distributions on antenna structures resulting from given excitations are generally unknown and very difficult to determine. In fact. the geometrically simple case of a straight conducting wire (linear antenna) excited by a voltage source in the middlet ij;ls hccn a sut?jcct o!'c.u!cnsivc ~.ci,::~t.cll li,r rll:llly y ~ 1 1 . 5 , ; l r lc l 1111: cx:ict c11;11.pc :~ntl currcnt distributions o n ;I wirc: of :I lini~c ~ ~ c l i u s ;lrc cxtrcnicly cotnplicatctl cvcn when the wire is assumed to be perfectly conducting. Fortunately. thc radiation licld of such an antenna is relatively insensitive to slight deviations in the current distribution, and a physically plausibie approximate current on the wire yields useful results for ncarly all practical purposes. Wc will exanline the radiation properties of linear antennas with assumed currents.

Documents

Cheng Emag